[ VIGRA Homepage | Function Index | Class Index | Namespaces | File List | Main Page ]

fftw3.hxx
1/************************************************************************/
2/* */
3/* Copyright 1998-2004 by Ullrich Koethe */
4/* */
5/* This file is part of the VIGRA computer vision library. */
6/* The VIGRA Website is */
7/* http://hci.iwr.uni-heidelberg.de/vigra/ */
8/* Please direct questions, bug reports, and contributions to */
9/* ullrich.koethe@iwr.uni-heidelberg.de or */
10/* vigra@informatik.uni-hamburg.de */
11/* */
12/* Permission is hereby granted, free of charge, to any person */
13/* obtaining a copy of this software and associated documentation */
14/* files (the "Software"), to deal in the Software without */
15/* restriction, including without limitation the rights to use, */
16/* copy, modify, merge, publish, distribute, sublicense, and/or */
17/* sell copies of the Software, and to permit persons to whom the */
18/* Software is furnished to do so, subject to the following */
19/* conditions: */
20/* */
21/* The above copyright notice and this permission notice shall be */
22/* included in all copies or substantial portions of the */
23/* Software. */
24/* */
25/* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND */
26/* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES */
27/* OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND */
28/* NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT */
29/* HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, */
30/* WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING */
31/* FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR */
32/* OTHER DEALINGS IN THE SOFTWARE. */
33/* */
34/************************************************************************/
35
36#ifndef VIGRA_FFTW3_HXX
37#define VIGRA_FFTW3_HXX
38
39#include <cmath>
40#include <functional>
41#include <complex>
42#include "stdimage.hxx"
43#include "copyimage.hxx"
44#include "transformimage.hxx"
45#include "combineimages.hxx"
46#include "numerictraits.hxx"
47#include "imagecontainer.hxx"
48#include <fftw3.h>
49
50namespace vigra {
51
52typedef double fftw_real;
53
54template <class T>
55struct FFTWReal;
56
57template <>
58struct FFTWReal<fftw_complex>
59{
60 typedef double type;
61};
62
63template <>
64struct FFTWReal<fftwf_complex>
65{
66 typedef float type;
67};
68
69template <>
70struct FFTWReal<fftwl_complex>
71{
72 typedef long double type;
73};
74
75template <class T>
76struct FFTWReal2Complex;
77
78template <>
79struct FFTWReal2Complex<double>
80{
81 typedef fftw_complex type;
82 typedef fftw_plan plan_type;
83};
84
85template <>
86struct FFTWReal2Complex<float>
87{
88 typedef fftwf_complex type;
89 typedef fftwf_plan plan_type;
90};
91
92template <>
93struct FFTWReal2Complex<long double>
94{
95 typedef fftwl_complex type;
96 typedef fftwl_plan plan_type;
97};
98
99/********************************************************/
100/* */
101/* FFTWComplex */
102/* */
103/********************************************************/
104
105/** \brief Wrapper class for the FFTW complex types '<TT>fftw_complex</TT>'.
106
107 This class encapsulates the low-level complex number types provided by the
108 <a href="http://www.fftw.org/">FFTW Fast Fourier Transform</a> library (i.e.
109 '<TT>fftw_complex</TT>', '<TT>fftwf_complex</TT>', '<TT>fftwl_complex</TT>').
110 In particular, it provides constructors, member functions and
111 \ref FFTWComplexOperators "arithmetic operators" that make FFTW complex numbers
112 compatible with <tt>std::complex</tt>. In addition, the class defines
113 transformations to polar coordinates and \ref FFTWComplexAccessors "accessors".
114
115 FFTWComplex implements the concepts \ref AlgebraicField and
116 \ref DivisionAlgebra. The standard image types <tt>FFTWRealImage</tt>
117 and <tt>FFTWComplexImage</tt> are defined.
118
119 <b>See also:</b>
120 <ul>
121 <li> \ref FFTWComplexTraits
122 <li> \ref FFTWComplexOperators
123 <li> \ref FFTWComplexAccessors
124 </ul>
125
126 <b>\#include</b> <vigra/fftw3.hxx> (for FFTW 3) or<br>
127 <b>\#include</b> <vigra/fftw.hxx> (for deprecated FFTW 2)<br>
128 Namespace: vigra
129*/
130template <class Real = double>
132{
133 public:
134 /** The wrapped complex type
135 */
136 typedef typename FFTWReal2Complex<Real>::type complex_type;
137
138 /** The complex' component type, as defined in '<TT>fftw3.h</TT>'
139 */
140 typedef Real value_type;
141
142 /** reference type (result of operator[])
143 */
145
146 /** const reference type (result of operator[] const)
147 */
149
150 /** iterator type (result of begin() )
151 */
153
154 /** const iterator type (result of begin() const)
155 */
156 typedef value_type const * const_iterator;
157
158 /** The norm type (result of magnitude())
159 */
161
162 /** The squared norm type (result of squaredMagnitde())
163 */
165
166 /** Construct from real and imaginary part.
167 Default: 0.
168 */
169 FFTWComplex(value_type const & re = 0.0, value_type const & im = 0.0)
170 {
171 data_[0] = re;
172 data_[1] = im;
173 }
174
175 /** Copy constructor.
176 */
178 {
179 data_[0] = o.data_[0];
180 data_[1] = o.data_[1];
181 }
182
183 /** Copy constructor.
184 */
185 template <class U>
187 {
188 data_[0] = (Real)o.real();
189 data_[1] = (Real)o.imag();
190 }
191
192 /** Construct from plain <TT>fftw_complex</TT>.
193 */
195 {
196 data_[0] = (Real)o[0];
197 data_[1] = (Real)o[1];
198 }
199
200 /** Construct from plain <TT>fftwf_complex</TT>.
201 */
203 {
204 data_[0] = (Real)o[0];
205 data_[1] = (Real)o[1];
206 }
207
208 /** Construct from plain <TT>fftwl_complex</TT>.
209 */
211 {
212 data_[0] = (Real)o[0];
213 data_[1] = (Real)o[1];
214 }
215
216 /** Construct from std::complex.
217 */
218 template <class T>
219 FFTWComplex(std::complex<T> const & o)
220 {
221 data_[0] = (Real)o.real();
222 data_[1] = (Real)o.imag();
223 }
224
225 /** Construct from TinyVector.
226 */
227 template <class T>
229 {
230 data_[0] = (Real)o[0];
231 data_[1] = (Real)o[1];
232 }
233
234 /** Assignment.
235 */
237 {
238 data_[0] = o.data_[0];
239 data_[1] = o.data_[1];
240 return *this;
241 }
242
243 /** Assignment.
244 */
245 template <class U>
247 {
248 data_[0] = (Real)o.real();
249 data_[1] = (Real)o.imag();
250 return *this;
251 }
252
253 /** Assignment.
254 */
256 {
257 data_[0] = (Real)o[0];
258 data_[1] = (Real)o[1];
259 return *this;
260 }
261
262 /** Assignment.
263 */
265 {
266 data_[0] = (Real)o[0];
267 data_[1] = (Real)o[1];
268 return *this;
269 }
270
271 /** Assignment.
272 */
274 {
275 data_[0] = (Real)o[0];
276 data_[1] = (Real)o[1];
277 return *this;
278 }
279
280 /** Assignment.
281 */
283 {
284 data_[0] = (Real)o;
285 data_[1] = 0.0;
286 return *this;
287 }
288
289 /** Assignment.
290 */
292 {
293 data_[0] = (Real)o;
294 data_[1] = 0.0;
295 return *this;
296 }
297
298 /** Assignment.
299 */
300 FFTWComplex& operator=(long double o)
301 {
302 data_[0] = (Real)o;
303 data_[1] = 0.0;
304 return *this;
305 }
306
307 /** Assignment.
308 */
309 template <class T>
311 {
312 data_[0] = (Real)o[0];
313 data_[1] = (Real)o[1];
314 return *this;
315 }
316
317 /** Assignment.
318 */
319 template <class T>
320 FFTWComplex& operator=(std::complex<T> const & o)
321 {
322 data_[0] = (Real)o.real();
323 data_[1] = (Real)o.imag();
324 return *this;
325 }
326
327 reference re()
328 { return data_[0]; }
329
330 const_reference re() const
331 { return data_[0]; }
332
333 reference real()
334 { return data_[0]; }
335
336 const_reference real() const
337 { return data_[0]; }
338
339 reference im()
340 { return data_[1]; }
341
342 const_reference im() const
343 { return data_[1]; }
344
345 reference imag()
346 { return data_[1]; }
347
348 const_reference imag() const
349 { return data_[1]; }
350
351 /** Unary negation.
352 */
354 { return FFTWComplex(-data_[0], -data_[1]); }
355
356 /** Squared magnitude x*conj(x)
357 */
359 { return data_[0]*data_[0]+data_[1]*data_[1]; }
360
361 /** Magnitude (length of radius vector).
362 */
364 { return VIGRA_CSTD::sqrt(squaredMagnitude()); }
365
366 /** Phase angle.
367 */
369 { return VIGRA_CSTD::atan2(data_[1], data_[0]); }
370
371 /** Access components as if number were a vector.
372 */
374 { return data_[i]; }
375
376 /** Read components as if number were a vector.
377 */
379 { return data_[i]; }
380
381 /** Length of complex number (always 2).
382 */
383 int size() const
384 { return 2; }
385
386 iterator begin()
387 { return data_; }
388
389 iterator end()
390 { return data_ + 2; }
391
392 const_iterator begin() const
393 { return data_; }
394
395 const_iterator end() const
396 { return data_ + 2; }
397
398 private:
399 complex_type data_;
400};
401
402/********************************************************/
403/* */
404/* FFTWComplexTraits */
405/* */
406/********************************************************/
407
408/** \page FFTWComplexTraits Numeric and Promote Traits of FFTWComplex
409
410 The numeric and promote traits for fftw_complex and FFTWComplex follow
411 the general specifications for \ref NumericPromotionTraits and
412 \ref AlgebraicField. They are explicitly specialized for the types
413 involved:
414
415 \code
416
417 template<>
418 struct NumericTraits<fftw_complex>
419 {
420 typedef fftw_complex Promote;
421 typedef fftw_complex RealPromote;
422 typedef fftw_complex ComplexPromote;
423 typedef fftw_real ValueType;
424
425 typedef VigraFalseType isIntegral;
426 typedef VigraFalseType isScalar;
427 typedef VigraFalseType isOrdered;
428 typedef VigraTrueType isComplex;
429
430 // etc.
431 };
432
433 template<class Real>
434 struct NumericTraits<FFTWComplex<Real> >
435 {
436 typedef FFTWComplex<Real> Promote;
437 typedef FFTWComplex<Real> RealPromote;
438 typedef FFTWComplex<Real> ComplexPromote;
439 typedef Real ValueType;
440
441 typedef VigraFalseType isIntegral;
442 typedef VigraFalseType isScalar;
443 typedef VigraFalseType isOrdered;
444 typedef VigraTrueType isComplex;
445
446 // etc.
447 };
448
449 template<>
450 struct NormTraits<fftw_complex>
451 {
452 typedef fftw_complex Type;
453 typedef fftw_real SquaredNormType;
454 typedef fftw_real NormType;
455 };
456
457 template<class Real>
458 struct NormTraits<FFTWComplex>
459 {
460 typedef FFTWComplex Type;
461 typedef fftw_real SquaredNormType;
462 typedef fftw_real NormType;
463 };
464
465 template <>
466 struct PromoteTraits<fftw_complex, fftw_complex>
467 {
468 typedef fftw_complex Promote;
469 };
470
471 template <>
472 struct PromoteTraits<fftw_complex, double>
473 {
474 typedef fftw_complex Promote;
475 };
476
477 template <>
478 struct PromoteTraits<double, fftw_complex>
479 {
480 typedef fftw_complex Promote;
481 };
482
483 template <class Real>
484 struct PromoteTraits<FFTWComplex<Real>, FFTWComplex<Real> >
485 {
486 typedef FFTWComplex<Real> Promote;
487 };
488
489 template <class Real>
490 struct PromoteTraits<FFTWComplex<Real>, double>
491 {
492 typedef FFTWComplex<Real> Promote;
493 };
494
495 template <class Real>
496 struct PromoteTraits<double, FFTWComplex<Real> >
497 {
498 typedef FFTWComplex<Real> Promote;
499 };
500 \endcode
501
502 <b>\#include</b> <vigra/fftw3.hxx> (for FFTW 3) or<br>
503 <b>\#include</b> <vigra/fftw.hxx> (for deprecated FFTW 2)<br>
504 Namespace: vigra
505
506*/
507template<>
508struct NumericTraits<fftw_complex>
509{
510 typedef fftw_complex Type;
511 typedef fftw_complex Promote;
512 typedef fftw_complex RealPromote;
513 typedef fftw_complex ComplexPromote;
514 typedef fftw_real ValueType;
515
516 typedef VigraFalseType isIntegral;
517 typedef VigraFalseType isScalar;
518 typedef NumericTraits<fftw_real>::isSigned isSigned;
519 typedef VigraFalseType isOrdered;
520 typedef VigraTrueType isComplex;
521
522 static FFTWComplex<> zero() { return FFTWComplex<>(0.0, 0.0); }
523 static FFTWComplex<> one() { return FFTWComplex<>(1.0, 0.0); }
524 static FFTWComplex<> nonZero() { return one(); }
525
526 static const Promote & toPromote(const Type & v) { return v; }
527 static const RealPromote & toRealPromote(const Type & v) { return v; }
528 static const Type & fromPromote(const Promote & v) { return v; }
529 static const Type & fromRealPromote(const RealPromote & v) { return v; }
530};
531
532template<class Real>
533struct NumericTraits<FFTWComplex<Real> >
534{
535 typedef FFTWComplex<Real> Type;
536 typedef FFTWComplex<Real> Promote;
537 typedef FFTWComplex<Real> RealPromote;
538 typedef FFTWComplex<Real> ComplexPromote;
539 typedef typename Type::value_type ValueType;
540
541 typedef VigraFalseType isIntegral;
542 typedef VigraFalseType isScalar;
543 typedef typename NumericTraits<ValueType>::isSigned isSigned;
544 typedef VigraFalseType isOrdered;
545 typedef VigraTrueType isComplex;
546
547 static Type zero() { return Type(0.0, 0.0); }
548 static Type one() { return Type(1.0, 0.0); }
549 static Type nonZero() { return one(); }
550
551 static const Promote & toPromote(const Type & v) { return v; }
552 static const RealPromote & toRealPromote(const Type & v) { return v; }
553 static const Type & fromPromote(const Promote & v) { return v; }
554 static const Type & fromRealPromote(const RealPromote & v) { return v; }
555};
556
557template<>
558struct NormTraits<fftw_complex>
559{
560 typedef fftw_complex Type;
561 typedef fftw_real SquaredNormType;
562 typedef fftw_real NormType;
563};
564
565template<class Real>
566struct NormTraits<FFTWComplex<Real> >
567{
568 typedef FFTWComplex<Real> Type;
569 typedef typename Type::SquaredNormType SquaredNormType;
570 typedef typename Type::NormType NormType;
571};
572
573template <>
574struct PromoteTraits<fftw_complex, fftw_complex>
575{
576 typedef fftw_complex Promote;
577};
578
579template <>
580struct PromoteTraits<fftw_complex, double>
581{
582 typedef fftw_complex Promote;
583};
584
585template <>
586struct PromoteTraits<double, fftw_complex>
587{
588 typedef fftw_complex Promote;
589};
590
591template <class Real>
592struct PromoteTraits<FFTWComplex<Real>, FFTWComplex<Real> >
593{
594 typedef FFTWComplex<Real> Promote;
595};
596
597template <class Real>
598struct PromoteTraits<FFTWComplex<Real>, double>
599{
600 typedef FFTWComplex<Real> Promote;
601};
602
603template <class Real>
604struct PromoteTraits<double, FFTWComplex<Real> >
605{
606 typedef FFTWComplex<Real> Promote;
607};
608
609template<class T>
610struct CanSkipInitialization<std::complex<T> >
611{
612 typedef typename CanSkipInitialization<T>::type type;
613 static const bool value = type::asBool;
614};
615
616template<class Real>
617struct CanSkipInitialization<FFTWComplex<Real> >
618{
619 typedef typename CanSkipInitialization<Real>::type type;
620 static const bool value = type::asBool;
621};
622
623namespace multi_math {
624
625template <class ARG>
626struct MultiMathOperand;
627
628template <class Real>
629struct MultiMathOperand<FFTWComplex<Real> >
630{
631 typedef MultiMathOperand<FFTWComplex<Real> > AllowOverload;
632 typedef FFTWComplex<Real> result_type;
633
634 static const int ndim = 0;
635
636 MultiMathOperand(FFTWComplex<Real> const & v)
637 : v_(v)
638 {}
639
640 template <class SHAPE>
641 bool checkShape(SHAPE const &) const
642 {
643 return true;
644 }
645
646 template <class SHAPE>
647 FFTWComplex<Real> const & operator[](SHAPE const &) const
648 {
649 return v_;
650 }
651
652 void inc(unsigned int /*LEVEL*/) const
653 {}
654
655 void reset(unsigned int /*LEVEL*/) const
656 {}
657
658 FFTWComplex<Real> const & operator*() const
659 {
660 return v_;
661 }
662
664};
665
666} // namespace multi_math
667
668template<class Ty>
669class FFTWAllocator
670{
671 public:
672 typedef std::size_t size_type;
673 typedef std::ptrdiff_t difference_type;
674 typedef Ty *pointer;
675 typedef const Ty *const_pointer;
676 typedef Ty& reference;
677 typedef const Ty& const_reference;
678 typedef Ty value_type;
679
680 pointer address(reference val) const
681 { return &val; }
682
683 const_pointer address(const_reference val) const
684 { return &val; }
685
686 template<class Other>
687 struct rebind
688 {
689 typedef FFTWAllocator<Other> other;
690 };
691
692 FFTWAllocator() throw()
693 {}
694
695 template<class Other>
696 FFTWAllocator(const FFTWAllocator<Other>& /*right*/) throw()
697 {}
698
699 template<class Other>
700 FFTWAllocator& operator=(const FFTWAllocator<Other>& /*right*/)
701 {
702 return *this;
703 }
704
705 pointer allocate(size_type count, void * = 0)
706 {
707 return (pointer)fftw_malloc(count * sizeof(Ty));
708 }
709
710 void deallocate(pointer ptr, size_type /*count*/)
711 {
712 fftw_free(ptr);
713 }
714
715 void construct(pointer ptr, const Ty& val)
716 {
717 new(ptr) Ty(val);
718
719 }
720
721 void destroy(pointer ptr)
722 {
723 ptr->~Ty();
724 }
725
726 size_type max_size() const throw()
727 {
728 return NumericTraits<std::ptrdiff_t>::max() / sizeof(Ty);
729 }
730};
731
732} // namespace vigra
733
734namespace std {
735
736template<class Real>
737class allocator<vigra::FFTWComplex<Real> >
738{
739 public:
740 typedef vigra::FFTWComplex<Real> value_type;
741 typedef size_t size_type;
742 typedef std::ptrdiff_t difference_type;
743 typedef value_type *pointer;
744 typedef const value_type *const_pointer;
745 typedef value_type& reference;
746 typedef const value_type& const_reference;
747
748 pointer address(reference val) const
749 { return &val; }
750
751 const_pointer address(const_reference val) const
752 { return &val; }
753
754 template<class Other>
755 struct rebind
756 {
757 typedef allocator<Other> other;
758 };
759
760 allocator() throw()
761 {}
762
763 template<class Other>
764 allocator(const allocator<Other>& /*right*/) throw()
765 {}
766
767 template<class Other>
768 allocator& operator=(const allocator<Other>& /*right*/)
769 {
770 return *this;
771 }
772
773 pointer allocate(size_type count, void * = 0)
774 {
775 return (pointer)fftw_malloc(count * sizeof(value_type));
776 }
777
778 void deallocate(pointer ptr, size_type /*count*/)
779 {
780 fftw_free(ptr);
781 }
782
783 void construct(pointer ptr, const value_type& val)
784 {
785 new(ptr) value_type(val);
786
787 }
788
789 void destroy(pointer ptr)
790 {
791 ptr->~value_type();
792 }
793
794 size_type max_size() const throw()
795 {
796 return vigra::NumericTraits<std::ptrdiff_t>::max() / sizeof(value_type);
797 }
798};
799
800} // namespace std
801
802namespace vigra {
803
804/********************************************************/
805/* */
806/* FFTWComplex Operations */
807/* */
808/********************************************************/
809
810/** \addtogroup FFTWComplexOperators Functions for FFTWComplex
811
812 <b>\#include</b> <vigra/fftw3.hxx> (for FFTW 3) or<br>
813 <b>\#include</b> <vigra/fftw.hxx> (for deprecated FFTW 2)<br>
814
815 These functions fulfill the requirements of an Algebraic Field.
816 Return types are determined according to \ref FFTWComplexTraits.
817
818 Namespace: vigra
819 <p>
820
821 */
822//@{
823 /// equal
824template <class R>
825inline bool operator ==(FFTWComplex<R> const &a, const FFTWComplex<R> &b) {
826 return a.re() == b.re() && a.im() == b.im();
827}
828
829template <class R>
830inline bool operator ==(FFTWComplex<R> const &a, double b) {
831 return a.re() == b && a.im() == 0.0;
832}
833
834template <class R>
835inline bool operator ==(double a, const FFTWComplex<R> &b) {
836 return a == b.re() && 0.0 == b.im();
837}
838
839 /// not equal
840template <class R>
841inline bool operator !=(FFTWComplex<R> const &a, const FFTWComplex<R> &b) {
842 return a.re() != b.re() || a.im() != b.im();
843}
844
845 /// not equal
846template <class R>
847inline bool operator !=(FFTWComplex<R> const &a, double b) {
848 return a.re() != b || a.im() != 0.0;
849}
850
851 /// not equal
852template <class R>
853inline bool operator !=(double a, const FFTWComplex<R> &b) {
854 return a != b.re() || 0.0 != b.im();
855}
856
857 /// add-assignment
858template <class R>
860 a.re() += b.re();
861 a.im() += b.im();
862 return a;
863}
864
865 /// subtract-assignment
866template <class R>
868 a.re() -= b.re();
869 a.im() -= b.im();
870 return a;
871}
872
873 /// multiply-assignment
874template <class R>
876 typename FFTWComplex<R>::value_type t = a.re()*b.re()-a.im()*b.im();
877 a.im() = a.re()*b.im()+a.im()*b.re();
878 a.re() = t;
879 return a;
880}
881
882 /// divide-assignment
883template <class R>
886 typename FFTWComplex<R>::value_type t = (a.re()*b.re()+a.im()*b.im())/sm;
887 a.im() = (b.re()*a.im()-a.re()*b.im())/sm;
888 a.re() = t;
889 return a;
890}
891
892 /// add-assignment with scalar double
893template <class R>
895 a.re() += (R)b;
896 return a;
897}
898
899 /// subtract-assignment with scalar double
900template <class R>
902 a.re() -= (R)b;
903 return a;
904}
905
906 /// multiply-assignment with scalar double
907template <class R>
909 a.re() *= (R)b;
910 a.im() *= (R)b;
911 return a;
912}
913
914 /// divide-assignment with scalar double
915template <class R>
917 a.re() /= (R)b;
918 a.im() /= (R)b;
919 return a;
920}
921
922 /// addition
923template <class R>
925 a += b;
926 return a;
927}
928
929 /// right addition with scalar double
930template <class R>
932 a += b;
933 return a;
934}
935
936 /// left addition with scalar double
937template <class R>
939 b += a;
940 return b;
941}
942
943 /// subtraction
944template <class R>
946 a -= b;
947 return a;
948}
949
950 /// right subtraction with scalar double
951template <class R>
953 a -= b;
954 return a;
955}
956
957 /// left subtraction with scalar double
958template <class R>
959inline FFTWComplex<R> operator -(double a, FFTWComplex<R> const & b) {
960 return (-b) += a;
961}
962
963 /// multiplication
964template <class R>
966 a *= b;
967 return a;
968}
969
970 /// right multiplication with scalar double
971template <class R>
973 a *= b;
974 return a;
975}
976
977 /// left multiplication with scalar double
978template <class R>
980 b *= a;
981 return b;
982}
983
984 /// division
985template <class R>
987 a /= b;
988 return a;
989}
990
991 /// right division with scalar double
992template <class R>
994 a /= b;
995 return a;
996}
997
998using VIGRA_CSTD::abs;
999
1000 /// absolute value (= magnitude)
1001template <class R>
1002inline typename FFTWComplex<R>::NormType abs(const FFTWComplex<R> &a)
1003{
1004 return a.magnitude();
1005}
1006
1007 /// phase
1008template <class R>
1009inline R arg(const FFTWComplex<R> &a)
1010{
1011 return a.phase();
1012}
1013
1014 /// real part
1015template <class R>
1016inline R real(const FFTWComplex<R> &a)
1017{
1018 return a.real();
1019}
1020
1021 /// imaginary part
1022template <class R>
1023inline R imag(const FFTWComplex<R> &a)
1024{
1025 return a.imag();
1026}
1027
1028 /// complex conjugate
1029template <class R>
1031{
1032 return FFTWComplex<R>(a.re(), -a.im());
1033}
1034
1035 /// norm (= magnitude)
1036template <class R>
1038{
1039 return a.magnitude();
1040}
1041
1042 /// squared norm (= squared magnitude)
1043template <class R>
1045{
1046 return a.squaredMagnitude();
1047}
1048
1049#define VIGRA_DEFINE_FFTW_COMPLEX_FUNCTION(fct) \
1050template <class R> \
1051inline FFTWComplex<R> fct(const FFTWComplex<R> &a) \
1052{ \
1053 return std::fct(reinterpret_cast<std::complex<R> const &>(a)); \
1054}
1055
1056VIGRA_DEFINE_FFTW_COMPLEX_FUNCTION(cos)
1057VIGRA_DEFINE_FFTW_COMPLEX_FUNCTION(cosh)
1058VIGRA_DEFINE_FFTW_COMPLEX_FUNCTION(exp)
1059VIGRA_DEFINE_FFTW_COMPLEX_FUNCTION(log)
1060VIGRA_DEFINE_FFTW_COMPLEX_FUNCTION(log10)
1061VIGRA_DEFINE_FFTW_COMPLEX_FUNCTION(sin)
1062VIGRA_DEFINE_FFTW_COMPLEX_FUNCTION(sinh)
1063VIGRA_DEFINE_FFTW_COMPLEX_FUNCTION(sqrt)
1064VIGRA_DEFINE_FFTW_COMPLEX_FUNCTION(tan)
1065VIGRA_DEFINE_FFTW_COMPLEX_FUNCTION(tanh)
1066
1067#undef VIGRA_DEFINE_FFTW_COMPLEX_FUNCTION
1068
1069template <class R>
1070inline FFTWComplex<R> pow(const FFTWComplex<R> &a, int e)
1071{
1072 return std::pow(reinterpret_cast<std::complex<R> const &>(a), e);
1073}
1074
1075template <class R>
1076inline FFTWComplex<R> pow(const FFTWComplex<R> &a, R const & e)
1077{
1078 return std::pow(reinterpret_cast<std::complex<R> const &>(a), e);
1079}
1080
1081template <class R>
1082inline FFTWComplex<R> pow(const FFTWComplex<R> &a, const FFTWComplex<R> & e)
1083{
1084 return std::pow(reinterpret_cast<std::complex<R> const &>(a),
1085 reinterpret_cast<std::complex<R> const &>(e));
1086}
1087
1088template <class R>
1089inline FFTWComplex<R> pow(R const & a, const FFTWComplex<R> &e)
1090{
1091 return std::pow(a, reinterpret_cast<std::complex<R> const &>(e));
1092}
1093
1094//@}
1095
1096} // namespace vigra
1097
1098namespace std {
1099
1100template <class Real>
1101ostream & operator<<(ostream & s, vigra::FFTWComplex<Real> const & v)
1102{
1103 s << std::complex<Real>(v.re(), v.im());
1104 return s;
1105}
1106
1107} // namespace std
1108
1109namespace vigra {
1110
1111/** \addtogroup StandardImageTypes
1112*/
1113//@{
1114
1115/********************************************************/
1116/* */
1117/* FFTWRealImage */
1118/* */
1119/********************************************************/
1120
1121 /** Float (<tt>fftw_real</tt>) image.
1122
1123 The type <tt>fftw_real</tt> is defined as <tt>double</tt> (in FFTW 2 it used to be
1124 either <tt>float</tt> or <tt>double</tt>, as specified during compilation of FFTW).
1125 FFTWRealImage uses \ref vigra::BasicImageIterator and \ref vigra::StandardAccessor and
1126 their const counterparts to access the data.
1127
1128 <b>\#include</b> <vigra/fftw3.hxx> (for FFTW 3) or<br>
1129 <b>\#include</b> <vigra/fftw.hxx> (for deprecated FFTW 2)<br>
1130 Namespace: vigra
1131 */
1133
1134/********************************************************/
1135/* */
1136/* FFTWComplexImage */
1137/* */
1138/********************************************************/
1139
1140template<class R>
1141struct IteratorTraits<
1143{
1144 typedef FFTWComplex<R> Type;
1145 typedef BasicImageIterator<Type, Type **> Iterator;
1146 typedef Iterator iterator;
1147 typedef BasicImageIterator<Type, Type **> mutable_iterator;
1148 typedef ConstBasicImageIterator<Type, Type **> const_iterator;
1149 typedef typename iterator::iterator_category iterator_category;
1150 typedef typename iterator::value_type value_type;
1151 typedef typename iterator::reference reference;
1152 typedef typename iterator::index_reference index_reference;
1153 typedef typename iterator::pointer pointer;
1154 typedef typename iterator::difference_type difference_type;
1155 typedef typename iterator::row_iterator row_iterator;
1156 typedef typename iterator::column_iterator column_iterator;
1157 typedef VectorAccessor<Type> default_accessor;
1158 typedef VectorAccessor<Type> DefaultAccessor;
1159 typedef VigraTrueType hasConstantStrides;
1160};
1161
1162template<class R>
1163struct IteratorTraits<
1164 ConstBasicImageIterator<FFTWComplex<R>, FFTWComplex<R> **> >
1165{
1166 typedef FFTWComplex<R> Type;
1167 typedef ConstBasicImageIterator<Type, Type **> Iterator;
1168 typedef Iterator iterator;
1169 typedef BasicImageIterator<Type, Type **> mutable_iterator;
1170 typedef ConstBasicImageIterator<Type, Type **> const_iterator;
1171 typedef typename iterator::iterator_category iterator_category;
1172 typedef typename iterator::value_type value_type;
1173 typedef typename iterator::reference reference;
1174 typedef typename iterator::index_reference index_reference;
1175 typedef typename iterator::pointer pointer;
1176 typedef typename iterator::difference_type difference_type;
1177 typedef typename iterator::row_iterator row_iterator;
1178 typedef typename iterator::column_iterator column_iterator;
1179 typedef VectorAccessor<Type> default_accessor;
1180 typedef VectorAccessor<Type> DefaultAccessor;
1181 typedef VigraTrueType hasConstantStrides;
1182};
1183
1184 /** Complex (FFTWComplex) image.
1185 It uses \ref vigra::BasicImageIterator and \ref vigra::StandardAccessor and
1186 their const counterparts to access the data.
1187
1188 <b>\#include</b> <vigra/fftw3.hxx> (for FFTW 3) or<br>
1189 <b>\#include</b> <vigra/fftw.hxx> (for deprecated FFTW 2)<br>
1190 Namespace: vigra
1191 */
1193
1194//@}
1195
1196/********************************************************/
1197/* */
1198/* FFTWComplex-Accessors */
1199/* */
1200/********************************************************/
1201
1202/** \addtogroup DataAccessors
1203*/
1204//@{
1205/** \defgroup FFTWComplexAccessors Accessors for FFTWComplex
1206
1207 Encapsulate access to pixels of type FFTWComplex
1208*/
1209//@{
1210 /** Encapsulate access to the the real part of a complex number.
1211
1212 <b>\#include</b> <vigra/fftw3.hxx> (for FFTW 3) or<br>
1213 <b>\#include</b> <vigra/fftw.hxx> (for deprecated FFTW 2)<br>
1214 Namespace: vigra
1215 */
1216template <class Real = double>
1218{
1219 public:
1220
1221 /// The accessor's value type.
1222 typedef Real value_type;
1223
1224 /// Read real part at iterator position.
1225 template <class ITERATOR>
1227 return (*i).re();
1228 }
1229
1230 /// Read real part at offset from iterator position.
1231 template <class ITERATOR, class DIFFERENCE>
1233 return i[d].re();
1234 }
1235
1236 /// Write real part at iterator position from a scalar.
1237 template <class ITERATOR>
1238 void set(value_type const & v, ITERATOR const & i) const {
1239 (*i).re()= v;
1240 }
1241
1242 /// Write real part at offset from iterator position from a scalar.
1243 template <class ITERATOR, class DIFFERENCE>
1244 void set(value_type const & v, ITERATOR const & i, DIFFERENCE d) const {
1245 i[d].re()= v;
1246 }
1247
1248 /// Write real part at iterator position into a scalar.
1249 template <class R, class ITERATOR>
1250 void set(FFTWComplex<R> const & v, ITERATOR const & i) const {
1251 *i = v.re();
1252 }
1253
1254 /// Write real part at offset from iterator position into a scalar.
1255 template <class R, class ITERATOR, class DIFFERENCE>
1256 void set(FFTWComplex<R> const & v, ITERATOR const & i, DIFFERENCE d) const {
1257 i[d] = v.re();
1258 }
1259};
1260
1261 /** Encapsulate access to the the imaginary part of a complex number.
1262
1263 <b>\#include</b> <vigra/fftw3.hxx> (for FFTW 3) or<br>
1264 <b>\#include</b> <vigra/fftw.hxx> (for deprecated FFTW 2)<br>
1265 Namespace: vigra
1266 */
1267template <class Real = double>
1269{
1270 public:
1271 /// The accessor's value type.
1272 typedef Real value_type;
1273
1274 /// Read imaginary part at iterator position.
1275 template <class ITERATOR>
1277 return (*i).im();
1278 }
1279
1280 /// Read imaginary part at offset from iterator position.
1281 template <class ITERATOR, class DIFFERENCE>
1283 return i[d].im();
1284 }
1285
1286 /// Write imaginary part at iterator position from a scalar.
1287 template <class ITERATOR>
1288 void set(value_type const & v, ITERATOR const & i) const {
1289 (*i).im()= v;
1290 }
1291
1292 /// Write imaginary part at offset from iterator position from a scalar.
1293 template <class ITERATOR, class DIFFERENCE>
1294 void set(value_type const & v, ITERATOR const & i, DIFFERENCE d) const {
1295 i[d].im()= v;
1296 }
1297
1298 /// Write imaginary part at iterator position into a scalar.
1299 template <class R, class ITERATOR>
1300 void set(FFTWComplex<R> const & v, ITERATOR const & i) const {
1301 *i = v.im();
1302 }
1303
1304 /// Write imaginary part at offset from iterator position into a scalar.
1305 template <class R, class ITERATOR, class DIFFERENCE>
1306 void set(FFTWComplex<R> const & v, ITERATOR const & i, DIFFERENCE d) const {
1307 i[d] = v.im();
1308 }
1309};
1310
1311 /** Write a real number into a complex one. The imaginary part is set
1312 to 0.
1313
1314 <b>\#include</b> <vigra/fftw3.hxx> (for FFTW 3) or<br>
1315 <b>\#include</b> <vigra/fftw.hxx> (for deprecated FFTW 2)<br>
1316 Namespace: vigra
1317 */
1318template <class Real = double>
1320: public FFTWRealAccessor<Real>
1321{
1322 public:
1323 /// The accessor's value type.
1324 typedef Real value_type;
1325
1326 /** Write real number at iterator position. Set imaginary part
1327 to 0.
1328 */
1329 template <class ITERATOR>
1330 void set(value_type const & v, ITERATOR const & i) const {
1331 (*i).re()= v;
1332 (*i).im()= 0;
1333 }
1334
1335 /** Write real number at offset from iterator position. Set imaginary part
1336 to 0.
1337 */
1338 template <class ITERATOR, class DIFFERENCE>
1339 void set(value_type const & v, ITERATOR const & i, DIFFERENCE d) const {
1340 i[d].re()= v;
1341 i[d].im()= 0;
1342 }
1343};
1344
1345 /** Calculate squared magnitude of complex number on the fly.
1346
1347 <b>\#include</b> <vigra/fftw3.hxx> (for FFTW 3) or<br>
1348 Namespace: vigra
1349 */
1350template <class Real = double>
1352{
1353 public:
1354 /// The accessor's value type.
1355 typedef Real value_type;
1356
1357 /// Read squared magnitude at iterator position.
1358 template <class ITERATOR>
1360 return (*i).squaredMagnitude();
1361 }
1362
1363 /// Read squared magnitude at offset from iterator position.
1364 template <class ITERATOR, class DIFFERENCE>
1366 return (i[d]).squaredMagnitude();
1367 }
1368};
1369
1370 /** Calculate magnitude of complex number on the fly.
1371
1372 <b>\#include</b> <vigra/fftw3.hxx> (for FFTW 3) or<br>
1373 <b>\#include</b> <vigra/fftw.hxx> (for deprecated FFTW 2)<br>
1374 Namespace: vigra
1375 */
1376template <class Real = double>
1378{
1379 public:
1380 /// The accessor's value type.
1381 typedef Real value_type;
1382
1383 /// Read magnitude at iterator position.
1384 template <class ITERATOR>
1386 return (*i).magnitude();
1387 }
1388
1389 /// Read magnitude at offset from iterator position.
1390 template <class ITERATOR, class DIFFERENCE>
1392 return (i[d]).magnitude();
1393 }
1394};
1395
1396 /** Calculate natural logarithm of magnitude of complex number on the fly.
1397
1398 <b>\#include</b> <vigra/fftw3.hxx> (for FFTW 3) or<br>
1399 <b>\#include</b> <vigra/fftw.hxx> (for deprecated FFTW 2)<br>
1400 Namespace: vigra
1401 */
1402template <class Real = double>
1404{
1405 public:
1406 /// The accessor's value type.
1407 typedef Real value_type;
1408
1409 /// Read natural log of magnitude at iterator position.
1410 template <class ITERATOR>
1412 return std::log((*i).magnitude() + 1);
1413 }
1414
1415 /// Read natural log of magnitude at offset from iterator position.
1416 template <class ITERATOR, class DIFFERENCE>
1418 return std::log((i[d]).magnitude() + 1);
1419 }
1420};
1421
1422 /** Calculate phase of complex number on the fly.
1423
1424 <b>\#include</b> <vigra/fftw3.hxx> (for FFTW 3) or<br>
1425 <b>\#include</b> <vigra/fftw.hxx> (for deprecated FFTW 2)<br>
1426 Namespace: vigra
1427 */
1428template <class Real = double>
1430{
1431 public:
1432 /// The accessor's value type.
1433 typedef Real value_type;
1434
1435 /// Read phase at iterator position.
1436 template <class ITERATOR>
1438 return (*i).phase();
1439 }
1440
1441 /// Read phase at offset from iterator position.
1442 template <class ITERATOR, class DIFFERENCE>
1444 return (i[d]).phase();
1445 }
1446};
1447
1448//@}
1449//@}
1450
1451/********************************************************/
1452/* */
1453/* Fourier Transform */
1454/* */
1455/********************************************************/
1456
1457/* \addtogroup FourierTransform Fast Fourier Transform
1458
1459 This documentation describes the VIGRA interface to FFTW version 3. The interface
1460 to the old FFTW version 2 (file "vigra/fftw.hxx") is deprecated.
1461
1462 VIGRA uses the <a href="http://www.fftw.org/">FFTW Fast Fourier
1463 Transform</a> package to perform Fourier transformations. VIGRA
1464 provides a wrapper for FFTW's complex number type (FFTWComplex),
1465 but FFTW's functions are used verbatim. If the image is stored as
1466 a FFTWComplexImage, the simplest call to an FFT function is like this:
1467
1468 \code
1469 vigra::FFTWComplexImage spatial(width,height), fourier(width,height);
1470 ... // fill image with data
1471
1472 // create a plan with estimated performance optimization
1473 fftw_plan forwardPlan = fftw_plan_dft_2d(height, width,
1474 (fftw_complex *)spatial.begin(), (fftw_complex *)fourier.begin(),
1475 FFTW_FORWARD, FFTW_ESTIMATE );
1476 // calculate FFT (this can be repeated as often as needed,
1477 // with fresh data written into the source array)
1478 fftw_execute(forwardPlan);
1479
1480 // release the plan memory
1481 fftw_destroy_plan(forwardPlan);
1482
1483 // likewise for the inverse transform
1484 fftw_plan backwardPlan = fftw_plan_dft_2d(height, width,
1485 (fftw_complex *)fourier.begin(), (fftw_complex *)spatial.begin(),
1486 FFTW_BACKWARD, FFTW_ESTIMATE);
1487 fftw_execute(backwardPlan);
1488 fftw_destroy_plan(backwardPlan);
1489
1490 // do not forget to normalize the result according to the image size
1491 transformImage(srcImageRange(spatial), destImage(spatial),
1492 std::bind(std::multiplies<FFTWComplex>(), 1.0 / width / height, std::placeholders::_1));
1493 \endcode
1494
1495 Note that in the creation of a plan, the height must be given
1496 first. Note also that <TT>spatial.begin()</TT> may only be passed
1497 to <TT>fftw_plan_dft_2d</TT> if the transform shall be applied to the
1498 entire image. When you want to restrict operation to an ROI, you
1499 can create a copy of the ROI in an image of appropriate size, or
1500 you may use the Guru interface to FFTW.
1501
1502 More information on using FFTW can be found <a href="http://www.fftw.org/doc/">here</a>.
1503
1504 FFTW produces fourier images that have the DC component (the
1505 origin of the Fourier space) in the upper left corner. Often, one
1506 wants the origin in the center of the image, so that frequencies
1507 always increase towards the border of the image. This can be
1508 achieved by calling \ref moveDCToCenter(). The inverse
1509 transformation is done by \ref moveDCToUpperLeft().
1510
1511 <b>\#include</b> <vigra/fftw3.hxx><br>
1512 Namespace: vigra
1513*/
1514
1515/** \addtogroup FourierTransform Fast Fourier Transform
1516
1517 VIGRA provides a powerful C++ API for the popular <a href="http://www.fftw.org/">FFTW library</a>
1518 for fast Fourier transforms. There are two versions of the API: an older one based on image
1519 iterators (and therefore restricted to 2D) and a new one based on \ref MultiArrayView that
1520 works for arbitrary dimensions. In addition, the functions \ref convolveFFT() and
1521 \ref applyFourierFilter() provide an easy-to-use interface for FFT-based convolution,
1522 a major application of Fourier transforms.
1523*/
1524//@{
1525
1526/********************************************************/
1527/* */
1528/* moveDCToCenter */
1529/* */
1530/********************************************************/
1531
1532/** \brief Rearrange the quadrants of a Fourier image so that the origin is
1533 in the image center.
1534
1535 FFTW produces fourier images where the DC component (origin of
1536 fourier space) is located in the upper left corner of the
1537 image. The quadrants are placed like this (using a 4x4 image for
1538 example):
1539
1540 \code
1541 DC 4 3 3
1542 4 4 3 3
1543 1 1 2 2
1544 1 1 2 2
1545 \endcode
1546
1547 After applying the function, the quadrants are at their usual places:
1548
1549 \code
1550 2 2 1 1
1551 2 2 1 1
1552 3 3 DC 4
1553 3 3 4 4
1554 \endcode
1555
1556 This transformation can be reversed by \ref moveDCToUpperLeft().
1557 Note that the 2D versions of this transformation must not be executed in place - input
1558 and output images must be different. In contrast, the nD version (with MultiArrayView
1559 argument) always works in-place.
1560
1561 <b> Declarations:</b>
1562
1563 use MultiArrayView (this works in-place, with arbitrary dimension N):
1564 \code
1565 namespace vigra {
1566 template <unsigned int N, class T, class Stride>
1567 void moveDCToCenter(MultiArrayView<N, T, Stride> a);
1568 }
1569 \endcode
1570
1571 pass iterators explicitly (2D only, not in-place):
1572 \code
1573 namespace vigra {
1574 template <class SrcImageIterator, class SrcAccessor,
1575 class DestImageIterator, class DestAccessor>
1576 void moveDCToCenter(SrcImageIterator src_upperleft,
1577 SrcImageIterator src_lowerright, SrcAccessor sa,
1578 DestImageIterator dest_upperleft, DestAccessor da);
1579 }
1580 \endcode
1581
1582
1583 use argument objects in conjunction with \ref ArgumentObjectFactories (2D only, not in-place):
1584 \code
1585 namespace vigra {
1586 template <class SrcImageIterator, class SrcAccessor,
1587 class DestImageIterator, class DestAccessor>
1588 void moveDCToCenter(
1589 triple<SrcImageIterator, SrcImageIterator, SrcAccessor> src,
1590 pair<DestImageIterator, DestAccessor> dest);
1591 }
1592 \endcode
1593
1594 <b> Usage:</b>
1595
1596 <b>\#include</b> <vigra/fftw3.hxx> (for 2D variants) <br>
1597 <b>\#include</b> <vigra/multi_fft.hxx> (for nD variants) <br>
1598 Namespace: vigra
1599
1600 \code
1601 vigra::FFTWComplexImage spatial(width,height), fourier(width,height);
1602 ... // fill image with data
1603
1604 // create a plan with estimated performance optimization
1605 fftw_plan forwardPlan = fftw_plan_dft_2d(height, width,
1606 (fftw_complex *)spatial.begin(), (fftw_complex *)fourier.begin(),
1607 FFTW_FORWARD, FFTW_ESTIMATE );
1608 // calculate FFT
1609 fftw_execute(forwardPlan);
1610
1611 vigra::FFTWComplexImage rearrangedFourier(width, height);
1612 moveDCToCenter(srcImageRange(fourier), destImage(rearrangedFourier));
1613
1614 // delete the plan
1615 fftw_destroy_plan(forwardPlan);
1616 \endcode
1617*/
1618doxygen_overloaded_function(template <...> void moveDCToCenter)
1619
1620template <class SrcImageIterator, class SrcAccessor,
1621 class DestImageIterator, class DestAccessor>
1625{
1626 int w = int(src_lowerright.x - src_upperleft.x);
1627 int h = int(src_lowerright.y - src_upperleft.y);
1628 int w1 = w/2;
1629 int h1 = h/2;
1630 int w2 = (w+1)/2;
1631 int h2 = (h+1)/2;
1632
1633 // 2. Quadrant zum 4.
1634 copyImage(srcIterRange(src_upperleft,
1635 src_upperleft + Diff2D(w2, h2), sa),
1636 destIter (dest_upperleft + Diff2D(w1, h1), da));
1637
1638 // 4. Quadrant zum 2.
1639 copyImage(srcIterRange(src_upperleft + Diff2D(w2, h2),
1641 destIter (dest_upperleft, da));
1642
1643 // 1. Quadrant zum 3.
1644 copyImage(srcIterRange(src_upperleft + Diff2D(w2, 0),
1645 src_upperleft + Diff2D(w, h2), sa),
1646 destIter (dest_upperleft + Diff2D(0, h1), da));
1647
1648 // 3. Quadrant zum 1.
1649 copyImage(srcIterRange(src_upperleft + Diff2D(0, h2),
1650 src_upperleft + Diff2D(w2, h), sa),
1651 destIter (dest_upperleft + Diff2D(w1, 0), da));
1652}
1653
1654template <class SrcImageIterator, class SrcAccessor,
1655 class DestImageIterator, class DestAccessor>
1656inline void moveDCToCenter(
1657 triple<SrcImageIterator, SrcImageIterator, SrcAccessor> src,
1658 pair<DestImageIterator, DestAccessor> dest)
1659{
1660 moveDCToCenter(src.first, src.second, src.third,
1661 dest.first, dest.second);
1662}
1663
1664/********************************************************/
1665/* */
1666/* moveDCToUpperLeft */
1667/* */
1668/********************************************************/
1669
1670/** \brief Rearrange the quadrants of a Fourier image so that the origin is
1671 in the image's upper left.
1672
1673 This function is the inversion of \ref moveDCToCenter(). See there
1674 for a detailed description and usage examples.
1675
1676 <b> Declarations:</b>
1677
1678 use MultiArrayView (this works in-place, with arbitrary dimension N):
1679 \code
1680 namespace vigra {
1681 template <unsigned int N, class T, class Stride>
1682 void moveDCToUpperLeft(MultiArrayView<N, T, Stride> a);
1683 }
1684 \endcode
1685
1686 pass iterators explicitly (2D only, not in-place):
1687 \code
1688 namespace vigra {
1689 template <class SrcImageIterator, class SrcAccessor,
1690 class DestImageIterator, class DestAccessor>
1691 void moveDCToUpperLeft(SrcImageIterator src_upperleft,
1692 SrcImageIterator src_lowerright, SrcAccessor sa,
1693 DestImageIterator dest_upperleft, DestAccessor da);
1694 }
1695 \endcode
1696
1697
1698 use argument objects in conjunction with \ref ArgumentObjectFactories (2D only, not in-place):
1699 \code
1700 namespace vigra {
1701 template <class SrcImageIterator, class SrcAccessor,
1702 class DestImageIterator, class DestAccessor>
1703 void moveDCToUpperLeft(
1704 triple<SrcImageIterator, SrcImageIterator, SrcAccessor> src,
1705 pair<DestImageIterator, DestAccessor> dest);
1706 }
1707 \endcode
1708*/
1709doxygen_overloaded_function(template <...> void moveDCToUpperLeft)
1710
1711template <class SrcImageIterator, class SrcAccessor,
1712 class DestImageIterator, class DestAccessor>
1716{
1717 int w = int(src_lowerright.x - src_upperleft.x);
1718 int h = int(src_lowerright.y - src_upperleft.y);
1719 int w2 = w/2;
1720 int h2 = h/2;
1721 int w1 = (w+1)/2;
1722 int h1 = (h+1)/2;
1723
1724 // 2. Quadrant zum 4.
1725 copyImage(srcIterRange(src_upperleft,
1726 src_upperleft + Diff2D(w2, h2), sa),
1727 destIter (dest_upperleft + Diff2D(w1, h1), da));
1728
1729 // 4. Quadrant zum 2.
1730 copyImage(srcIterRange(src_upperleft + Diff2D(w2, h2),
1732 destIter (dest_upperleft, da));
1733
1734 // 1. Quadrant zum 3.
1735 copyImage(srcIterRange(src_upperleft + Diff2D(w2, 0),
1736 src_upperleft + Diff2D(w, h2), sa),
1737 destIter (dest_upperleft + Diff2D(0, h1), da));
1738
1739 // 3. Quadrant zum 1.
1740 copyImage(srcIterRange(src_upperleft + Diff2D(0, h2),
1741 src_upperleft + Diff2D(w2, h), sa),
1742 destIter (dest_upperleft + Diff2D(w1, 0), da));
1743}
1744
1745template <class SrcImageIterator, class SrcAccessor,
1746 class DestImageIterator, class DestAccessor>
1747inline void moveDCToUpperLeft(
1748 triple<SrcImageIterator, SrcImageIterator, SrcAccessor> src,
1749 pair<DestImageIterator, DestAccessor> dest)
1750{
1751 moveDCToUpperLeft(src.first, src.second, src.third,
1752 dest.first, dest.second);
1753}
1754
1755namespace detail {
1756
1757template <class T>
1758void
1759fourierTransformImpl(FFTWComplexImage::const_traverser sul,
1760 FFTWComplexImage::const_traverser slr, FFTWComplexImage::ConstAccessor src,
1761 FFTWComplexImage::traverser dul, FFTWComplexImage::Accessor dest, T sign)
1762{
1763 int w = int(slr.x - sul.x);
1764 int h = int(slr.y - sul.y);
1765
1766 FFTWComplexImage sworkImage, dworkImage;
1767
1768 fftw_complex * srcPtr = (fftw_complex *)(&*sul);
1769 fftw_complex * destPtr = (fftw_complex *)(&*dul);
1770
1771 // test for right memory layout (fftw expects a 2*width*height floats array)
1772 if (h > 1 && &(*(sul + Diff2D(w, 0))) != &(*(sul + Diff2D(0, 1))))
1773 {
1774 sworkImage.resize(w, h);
1775 copyImage(srcIterRange(sul, slr, src), destImage(sworkImage));
1776 srcPtr = (fftw_complex *)(&(*sworkImage.upperLeft()));
1777 }
1778 if (h > 1 && &(*(dul + Diff2D(w, 0))) != &(*(dul + Diff2D(0, 1))))
1779 {
1780 dworkImage.resize(w, h);
1781 destPtr = (fftw_complex *)(&(*dworkImage.upperLeft()));
1782 }
1783
1784 fftw_plan plan = fftw_plan_dft_2d(h, w, srcPtr, destPtr, sign, FFTW_ESTIMATE );
1785 fftw_execute(plan);
1786 fftw_destroy_plan(plan);
1787
1788 if (h > 1 && &(*(dul + Diff2D(w, 0))) != &(*(dul + Diff2D(0, 1))))
1789 {
1790 copyImage(srcImageRange(dworkImage), destIter(dul, dest));
1791 }
1792}
1793
1794} // namespace detail
1795
1796/********************************************************/
1797/* */
1798/* fourierTransform */
1799/* */
1800/********************************************************/
1801
1802/** \brief Compute forward and inverse Fourier transforms.
1803
1804 The array referring to the spatial domain (i.e. the input in a forward transform,
1805 and the output in an inverse transform) may be scalar or complex. The array representing
1806 the frequency domain (i.e. output for forward transform, input for inverse transform)
1807 must always be complex.
1808
1809 The new implementations (those using MultiArrayView arguments) perform a normalized transform,
1810 whereas the old ones (using 2D iterators or argument objects) perform an un-normalized
1811 transform (i.e. the result of the inverse transform is scaled by the number of pixels).
1812
1813 In general, input and output arrays must have the same shape, with the exception of the
1814 special <a href="http://www.fftw.org/doc/Multi_002dDimensional-DFTs-of-Real-Data.html">R2C
1815 and C2R modes</a> defined by FFTW.
1816
1817 The R2C transform reduces the redundancy in the Fourier representation of a real-valued signal:
1818 Since the Fourier representation of a real signal is symmetric, about half of the Fourier coefficients
1819 can simply be dropped. By convention, this reduction is applied to the first (innermost) dimension,
1820 such that <tt>fourier.shape(0) == spatial.shape(0)/2 + 1</tt> holds. The correct frequency domain
1821 shape can be conveniently computed by means of the function \ref fftwCorrespondingShapeR2C().
1822
1823 Note that your program must always link against <tt>libfftw3</tt>. If you want to compute Fourier
1824 transforms for <tt>float</tt> or <tt>long double</tt> arrays, you must <i>additionally</i> link against <tt>libfftw3f</tt> and <tt>libfftw3l</tt> respectively. (Old-style functions only support <tt>double</tt>).
1825
1826 The Fourier transform functions internally create <a href="http://www.fftw.org/doc/Using-Plans.html">FFTW plans</a>
1827 which control the algorithm details. The plans are creates with the flag <tt>FFTW_ESTIMATE</tt>, i.e.
1828 optimal settings are guessed or read from saved "wisdom" files. If you need more control over planning,
1829 you can use the class \ref FFTWPlan.
1830
1831 <b> Declarations:</b>
1832
1833 use complex-valued MultiArrayView arguments (this works for arbitrary dimension N):
1834 \code
1835 namespace vigra {
1836 template <unsigned int N, class Real, class C1, class C2>
1837 void
1838 fourierTransform(MultiArrayView<N, FFTWComplex<Real>, C1> in,
1839 MultiArrayView<N, FFTWComplex<Real>, C2> out);
1840
1841 template <unsigned int N, class Real, class C1, class C2>
1842 void
1843 fourierTransformInverse(MultiArrayView<N, FFTWComplex<Real>, C1> in,
1844 MultiArrayView<N, FFTWComplex<Real>, C2> out);
1845 }
1846 \endcode
1847
1848 use real-valued MultiArrayView in the spatial domain, complex-valued MultiArrayView
1849 in the frequency domain (this works for arbitrary dimension N, and also supports
1850 the R2C and C2R transform, depending on the array shape in the frequency domain):
1851 \code
1852 namespace vigra {
1853 template <unsigned int N, class Real, class C1, class C2>
1854 void
1855 fourierTransform(MultiArrayView<N, Real, C1> in,
1856 MultiArrayView<N, FFTWComplex<Real>, C2> out);
1857
1858 template <unsigned int N, class Real, class C1, class C2>
1859 void
1860 fourierTransformInverse(MultiArrayView<N, FFTWComplex<Real>, C1> in,
1861 MultiArrayView<N, Real, C2> out);
1862 }
1863 \endcode
1864
1865 pass iterators explicitly (2D only, double only):
1866 \code
1867 namespace vigra {
1868 template <class SrcImageIterator, class SrcAccessor>
1869 void fourierTransform(SrcImageIterator srcUpperLeft,
1870 SrcImageIterator srcLowerRight, SrcAccessor src,
1871 FFTWComplexImage::traverser destUpperLeft, FFTWComplexImage::Accessor dest);
1872
1873 void
1874 fourierTransformInverse(FFTWComplexImage::const_traverser sul,
1875 FFTWComplexImage::const_traverser slr, FFTWComplexImage::ConstAccessor src,
1876 FFTWComplexImage::traverser dul, FFTWComplexImage::Accessor dest)
1877 }
1878 \endcode
1879
1880 use argument objects in conjunction with \ref ArgumentObjectFactories (2D only, double only):
1881 \code
1882 namespace vigra {
1883 template <class SrcImageIterator, class SrcAccessor>
1884 void fourierTransform(triple<SrcImageIterator, SrcImageIterator, SrcAccessor> src,
1885 pair<FFTWComplexImage::traverser, FFTWComplexImage::Accessor> dest);
1886
1887 void
1888 fourierTransformInverse(triple<FFTWComplexImage::const_traverser,
1889 FFTWComplexImage::const_traverser, FFTWComplexImage::ConstAccessor> src,
1890 pair<FFTWComplexImage::traverser, FFTWComplexImage::Accessor> dest);
1891 }
1892 \endcode
1893
1894 <b> Usage:</b>
1895
1896 <b>\#include</b> <vigra/fftw3.hxx> (old-style 2D variants)<br>
1897 <b>\#include</b> <vigra/multi_fft.hxx> (new-style nD variants)<br>
1898 Namespace: vigra
1899
1900 old-style example (using FFTWComplexImage):
1901 \code
1902 // compute complex Fourier transform of a real image, old-style implementation
1903 vigra::DImage src(w, h);
1904 vigra::FFTWComplexImage fourier(w, h);
1905
1906 fourierTransform(srcImageRange(src), destImage(fourier));
1907
1908 // compute inverse Fourier transform
1909 // note that both source and destination image must be of type vigra::FFTWComplexImage
1910 vigra::FFTWComplexImage inverseFourier(w, h);
1911
1912 fourierTransformInverse(srcImageRange(fourier), destImage(inverseFourier));
1913 \endcode
1914
1915 new-style examples (using MultiArray):
1916 \code
1917 // compute Fourier transform of a real array, using the R2C algorithm
1918 MultiArray<2, double> src(Shape2(w, h));
1919 MultiArray<2, FFTWComplex<double> > fourier(fftwCorrespondingShapeR2C(src.shape()));
1920
1921 fourierTransform(src, fourier);
1922
1923 // compute inverse Fourier transform, using the C2R algorithm
1924 MultiArray<2, double> dest(src.shape());
1925 fourierTransformInverse(fourier, dest);
1926 \endcode
1927
1928 \code
1929 // compute Fourier transform of a real array with standard algorithm
1930 MultiArray<2, double> src(Shape2(w, h));
1931 MultiArray<2, FFTWComplex<double> > fourier(src.shape());
1932
1933 fourierTransform(src, fourier);
1934
1935 // compute inverse Fourier transform, using the C2R algorithm
1936 MultiArray<2, double> dest(src.shape());
1937 fourierTransformInverse(fourier, dest);
1938 \endcode
1939 Complex input arrays are handled in the same way.
1940
1941*/
1942doxygen_overloaded_function(template <...> void fourierTransform)
1943
1944inline void
1945fourierTransform(FFTWComplexImage::const_traverser sul,
1946 FFTWComplexImage::const_traverser slr, FFTWComplexImage::ConstAccessor src,
1947 FFTWComplexImage::traverser dul, FFTWComplexImage::Accessor dest)
1948{
1949 detail::fourierTransformImpl(sul, slr, src, dul, dest, FFTW_FORWARD);
1950}
1951
1952template <class SrcImageIterator, class SrcAccessor>
1953void fourierTransform(SrcImageIterator srcUpperLeft,
1954 SrcImageIterator srcLowerRight, SrcAccessor sa,
1955 FFTWComplexImage::traverser destUpperLeft, FFTWComplexImage::Accessor da)
1956{
1957 // copy real input images into a complex one...
1958 int w= srcLowerRight.x - srcUpperLeft.x;
1959 int h= srcLowerRight.y - srcUpperLeft.y;
1960
1961 FFTWComplexImage workImage(w, h);
1962 copyImage(srcIterRange(srcUpperLeft, srcLowerRight, sa),
1963 destImage(workImage, FFTWWriteRealAccessor<>()));
1964
1965 // ...and call the complex -> complex version of the algorithm
1966 FFTWComplexImage const & cworkImage = workImage;
1967 fourierTransform(cworkImage.upperLeft(), cworkImage.lowerRight(), cworkImage.accessor(),
1968 destUpperLeft, da);
1969}
1970
1971template <class SrcImageIterator, class SrcAccessor>
1972inline
1973void fourierTransform(triple<SrcImageIterator, SrcImageIterator, SrcAccessor> src,
1974 pair<FFTWComplexImage::traverser, FFTWComplexImage::Accessor> dest)
1975{
1976 fourierTransform(src.first, src.second, src.third, dest.first, dest.second);
1977}
1978
1979/** \brief Compute inverse Fourier transforms.
1980
1981 See \ref fourierTransform() for details.
1982*/
1983doxygen_overloaded_function(template <...> void fourierTransformInverse)
1984
1985inline void
1986fourierTransformInverse(FFTWComplexImage::const_traverser sul,
1987 FFTWComplexImage::const_traverser slr, FFTWComplexImage::ConstAccessor src,
1988 FFTWComplexImage::traverser dul, FFTWComplexImage::Accessor dest)
1989{
1990 detail::fourierTransformImpl(sul, slr, src, dul, dest, FFTW_BACKWARD);
1991}
1992
1993template <class DestImageIterator, class DestAccessor>
1994void fourierTransformInverse(FFTWComplexImage::const_traverser sul,
1995 FFTWComplexImage::const_traverser slr, FFTWComplexImage::ConstAccessor src,
1996 DestImageIterator dul, DestAccessor dest)
1997{
1998 int w = slr.x - sul.x;
1999 int h = slr.y - sul.y;
2000
2001 FFTWComplexImage workImage(w, h);
2002 fourierTransformInverse(sul, slr, src, workImage.upperLeft(), workImage.accessor());
2003 copyImage(srcImageRange(workImage), destIter(dul, dest));
2004}
2005
2006
2007template <class DestImageIterator, class DestAccessor>
2008inline void
2009fourierTransformInverse(triple<FFTWComplexImage::const_traverser,
2010 FFTWComplexImage::const_traverser, FFTWComplexImage::ConstAccessor> src,
2011 pair<DestImageIterator, DestAccessor> dest)
2012{
2013 fourierTransformInverse(src.first, src.second, src.third, dest.first, dest.second);
2014}
2015
2016
2017
2018/********************************************************/
2019/* */
2020/* applyFourierFilter */
2021/* */
2022/********************************************************/
2023
2024/** \brief Apply a filter (defined in the frequency domain) to an image.
2025
2026 After transferring the image into the frequency domain, it is
2027 multiplied pixel-wise with the filter and transformed back. The
2028 result is put into the given destination image which must have the right size.
2029 The result will be normalized to compensate for the two FFTs.
2030
2031 If the destination image is scalar, only the real part of the result image is
2032 retained. In this case, you are responsible for choosing a filter image
2033 which ensures a zero imaginary part of the result (e.g. use a real, even symmetric
2034 filter image, or a purely imaginary, odd symmetric one).
2035
2036 The DC entry of the filter must be in the upper left, which is the
2037 position where FFTW expects it (see \ref moveDCToUpperLeft()).
2038
2039 See also \ref convolveFFT() for corresponding functionality on the basis of the
2040 \ref MultiArrayView interface.
2041
2042 <b> Declarations:</b>
2043
2044 pass 2D array views:
2045 \code
2046 namespace vigra {
2047 template <class SrcImageIterator, class SrcAccessor,
2048 class FilterImageIterator, class FilterAccessor,
2049 class DestImageIterator, class DestAccessor>
2050 void applyFourierFilter(SrcImageIterator srcUpperLeft,
2051 SrcImageIterator srcLowerRight, SrcAccessor sa,
2052 FilterImageIterator filterUpperLeft, FilterAccessor fa,
2053 DestImageIterator destUpperLeft, DestAccessor da);
2054 }
2055 \endcode
2056
2057 pass \ref ImageIterators and \ref DataAccessors :
2058 \code
2059 namespace vigra {
2060 template <class SrcImageIterator, class SrcAccessor,
2061 class FilterImageIterator, class FilterAccessor,
2062 class DestImageIterator, class DestAccessor>
2063 void applyFourierFilter(SrcImageIterator srcUpperLeft,
2064 SrcImageIterator srcLowerRight, SrcAccessor sa,
2065 FilterImageIterator filterUpperLeft, FilterAccessor fa,
2066 DestImageIterator destUpperLeft, DestAccessor da);
2067 }
2068 \endcode
2069
2070 use argument objects in conjunction with \ref ArgumentObjectFactories :
2071 \code
2072 namespace vigra {
2073 template <class SrcImageIterator, class SrcAccessor,
2074 class FilterImageIterator, class FilterAccessor,
2075 class DestImageIterator, class DestAccessor>
2076 void applyFourierFilter(triple<SrcImageIterator, SrcImageIterator, SrcAccessor> src,
2077 pair<FilterImageIterator, FilterAccessor> filter,
2078 pair<DestImageIterator, DestAccessor> dest);
2079 }
2080 \endcode
2081
2082 <b> Usage:</b>
2083
2084 <b>\#include</b> <vigra/fftw3.hxx><br>
2085 Namespace: vigra
2086
2087 \code
2088 // create a Gaussian filter in Fourier space
2089 vigra::FImage gaussFilter(w, h), filter(w, h);
2090 for(int y=0; y<h; ++y)
2091 for(int x=0; x<w; ++x)
2092 {
2093 xx = float(x - w / 2) / w;
2094 yy = float(y - h / 2) / h;
2095
2096 gaussFilter(x,y) = std::exp(-(xx*xx + yy*yy) / 2.0 * scale);
2097 }
2098
2099 // applyFourierFilter() expects the filter's DC in the upper left
2100 moveDCToUpperLeft(srcImageRange(gaussFilter), destImage(filter));
2101
2102 vigra::FFTWComplexImage result(w, h);
2103
2104 vigra::applyFourierFilter(srcImageRange(image), srcImage(filter), result);
2105 \endcode
2106
2107 For inspection of the result, \ref FFTWMagnitudeAccessor might be
2108 useful. If you want to apply the same filter repeatedly, it may be more
2109 efficient to use the FFTW functions directly with FFTW plans optimized
2110 for good performance.
2111*/
2112doxygen_overloaded_function(template <...> void applyFourierFilter)
2113
2114template <class SrcImageIterator, class SrcAccessor,
2116 class DestImageIterator, class DestAccessor>
2121{
2122 // copy real input images into a complex one...
2123 int w = int(srcLowerRight.x - srcUpperLeft.x);
2124 int h = int(srcLowerRight.y - srcUpperLeft.y);
2125
2127 copyImage(srcIterRange(srcUpperLeft, srcLowerRight, sa),
2128 destImage(workImage, FFTWWriteRealAccessor<>()));
2129
2130 // ...and call the impl
2132 applyFourierFilterImpl(cworkImage.upperLeft(), cworkImage.lowerRight(), cworkImage.accessor(),
2134 destUpperLeft, da);
2135}
2136
2137template <class FilterImageIterator, class FilterAccessor,
2138 class DestImageIterator, class DestAccessor>
2139inline
2141 FFTWComplexImage::const_traverser srcUpperLeft,
2142 FFTWComplexImage::const_traverser srcLowerRight,
2143 FFTWComplexImage::ConstAccessor sa,
2144 FilterImageIterator filterUpperLeft, FilterAccessor fa,
2145 DestImageIterator destUpperLeft, DestAccessor da)
2146{
2147 int w = srcLowerRight.x - srcUpperLeft.x;
2148 int h = srcLowerRight.y - srcUpperLeft.y;
2149
2150 // test for right memory layout (fftw expects a 2*width*height floats array)
2151 if (&(*(srcUpperLeft + Diff2D(w, 0))) == &(*(srcUpperLeft + Diff2D(0, 1))))
2152 applyFourierFilterImpl(srcUpperLeft, srcLowerRight, sa,
2153 filterUpperLeft, fa,
2154 destUpperLeft, da);
2155 else
2156 {
2157 FFTWComplexImage workImage(w, h);
2158 copyImage(srcIterRange(srcUpperLeft, srcLowerRight, sa),
2159 destImage(workImage));
2160
2161 FFTWComplexImage const & cworkImage = workImage;
2162 applyFourierFilterImpl(cworkImage.upperLeft(), cworkImage.lowerRight(), cworkImage.accessor(),
2163 filterUpperLeft, fa,
2164 destUpperLeft, da);
2165 }
2166}
2167
2168template <class SrcImageIterator, class SrcAccessor,
2169 class FilterImageIterator, class FilterAccessor,
2170 class DestImageIterator, class DestAccessor>
2171inline
2172void applyFourierFilter(triple<SrcImageIterator, SrcImageIterator, SrcAccessor> src,
2173 pair<FilterImageIterator, FilterAccessor> filter,
2174 pair<DestImageIterator, DestAccessor> dest)
2175{
2176 applyFourierFilter(src.first, src.second, src.third,
2177 filter.first, filter.second,
2178 dest.first, dest.second);
2179}
2180
2181template <class FilterImageIterator, class FilterAccessor,
2182 class DestImageIterator, class DestAccessor>
2183void applyFourierFilterImpl(
2184 FFTWComplexImage::const_traverser srcUpperLeft,
2185 FFTWComplexImage::const_traverser srcLowerRight,
2186 FFTWComplexImage::ConstAccessor,
2187 FilterImageIterator filterUpperLeft, FilterAccessor fa,
2188 DestImageIterator destUpperLeft, DestAccessor da)
2189{
2190 int w = int(srcLowerRight.x - srcUpperLeft.x);
2191 int h = int(srcLowerRight.y - srcUpperLeft.y);
2192
2193 FFTWComplexImage complexResultImg(srcLowerRight - srcUpperLeft);
2194
2195 // FFT from srcImage to complexResultImg
2196 fftw_plan forwardPlan=
2197 fftw_plan_dft_2d(h, w, (fftw_complex *)&(*srcUpperLeft),
2198 (fftw_complex *)complexResultImg.begin(),
2199 FFTW_FORWARD, FFTW_ESTIMATE );
2200 fftw_execute(forwardPlan);
2201 fftw_destroy_plan(forwardPlan);
2202
2203 // convolve in freq. domain (in complexResultImg)
2204 combineTwoImages(srcImageRange(complexResultImg), srcIter(filterUpperLeft, fa),
2205 destImage(complexResultImg), std::multiplies<FFTWComplex<> >());
2206
2207 // FFT back into spatial domain (inplace in complexResultImg)
2208 fftw_plan backwardPlan=
2209 fftw_plan_dft_2d(h, w, (fftw_complex *)complexResultImg.begin(),
2210 (fftw_complex *)complexResultImg.begin(),
2211 FFTW_BACKWARD, FFTW_ESTIMATE);
2212 fftw_execute(backwardPlan);
2213 fftw_destroy_plan(backwardPlan);
2214
2215 typedef typename
2216 NumericTraits<typename DestAccessor::value_type>::isScalar
2217 isScalarResult;
2218
2219 // normalization (after FFTs), maybe stripping imaginary part
2220 applyFourierFilterImplNormalization(complexResultImg, destUpperLeft, da,
2221 isScalarResult());
2222}
2223
2224template <class DestImageIterator, class DestAccessor>
2225void applyFourierFilterImplNormalization(FFTWComplexImage const &srcImage,
2226 DestImageIterator destUpperLeft,
2227 DestAccessor da,
2228 VigraFalseType)
2229{
2230 double normFactor= 1.0/(srcImage.width() * srcImage.height());
2231
2232 for(int y=0; y<srcImage.height(); y++, destUpperLeft.y++)
2233 {
2234 DestImageIterator dIt= destUpperLeft;
2235 for(int x= 0; x< srcImage.width(); x++, dIt.x++)
2236 {
2237 da.setComponent(srcImage(x, y).re()*normFactor, dIt, 0);
2238 da.setComponent(srcImage(x, y).im()*normFactor, dIt, 1);
2239 }
2240 }
2241}
2242
2243inline
2244void applyFourierFilterImplNormalization(FFTWComplexImage const & srcImage,
2245 FFTWComplexImage::traverser destUpperLeft,
2246 FFTWComplexImage::Accessor da,
2247 VigraFalseType)
2248{
2249 transformImage(srcImageRange(srcImage), destIter(destUpperLeft, da),
2250 linearIntensityTransform<FFTWComplex<> >(1.0/(srcImage.width() * srcImage.height())));
2251}
2252
2253template <class DestImageIterator, class DestAccessor>
2254void applyFourierFilterImplNormalization(FFTWComplexImage const & srcImage,
2255 DestImageIterator destUpperLeft,
2256 DestAccessor da,
2257 VigraTrueType)
2258{
2259 double normFactor= 1.0/(srcImage.width() * srcImage.height());
2260
2261 for(int y=0; y<srcImage.height(); y++, destUpperLeft.y++)
2262 {
2263 DestImageIterator dIt= destUpperLeft;
2264 for(int x= 0; x< srcImage.width(); x++, dIt.x++)
2265 da.set(srcImage(x, y).re()*normFactor, dIt);
2266 }
2267}
2268
2269/**********************************************************/
2270/* */
2271/* applyFourierFilterFamily */
2272/* */
2273/**********************************************************/
2274
2275/** \brief Apply an array of filters (defined in the frequency domain) to an image.
2276
2277 This provides the same functionality as \ref applyFourierFilter(),
2278 but applying several filters at once allows to avoid
2279 repeated Fourier transforms of the source image.
2280
2281 Filters and result images must be stored in \ref vigra::ImageArray data
2282 structures. In contrast to \ref applyFourierFilter(), this function adjusts
2283 the size of the result images and the the length of the array.
2284
2285 <b> Declarations:</b>
2286
2287 pass 2D array views:
2288 \code
2289 namespace vigra {
2290 template <class SrcImageIterator, class SrcAccessor, class FilterType>
2291 void applyFourierFilterFamily(SrcImageIterator srcUpperLeft,
2292 SrcImageIterator srcLowerRight, SrcAccessor sa,
2293 const ImageArray<FilterType> &filters,
2294 ImageArray<FFTWComplexImage> &results)
2295 }
2296 \endcode
2297
2298 pass \ref ImageIterators and \ref DataAccessors :
2299 \code
2300 namespace vigra {
2301 template <class SrcImageIterator, class SrcAccessor, class FilterType>
2302 void applyFourierFilterFamily(SrcImageIterator srcUpperLeft,
2303 SrcImageIterator srcLowerRight, SrcAccessor sa,
2304 const ImageArray<FilterType> &filters,
2305 ImageArray<FFTWComplexImage> &results)
2306 }
2307 \endcode
2308
2309 use argument objects in conjunction with \ref ArgumentObjectFactories :
2310 \code
2311 namespace vigra {
2312 template <class SrcImageIterator, class SrcAccessor, class FilterType>
2313 void applyFourierFilterFamily(triple<SrcImageIterator, SrcImageIterator, SrcAccessor> src,
2314 const ImageArray<FilterType> &filters,
2315 ImageArray<FFTWComplexImage> &results)
2316 }
2317 \endcode
2318
2319 <b> Usage:</b>
2320
2321 <b>\#include</b> <vigra/fftw3.hxx><br>
2322 Namespace: vigra
2323
2324 \code
2325 // assuming the presence of a real-valued image named "image" to
2326 // be filtered in this example
2327
2328 vigra::ImageArray<vigra::FImage> filters(16, image.size());
2329
2330 for (int i=0; i<filters.size(); i++)
2331 // create some meaningful filters here
2332 createMyFilterOfScale(i, destImage(filters[i]));
2333
2334 vigra::ImageArray<vigra::FFTWComplexImage> results();
2335
2336 vigra::applyFourierFilterFamily(srcImageRange(image), filters, results);
2337 \endcode
2338*/
2339doxygen_overloaded_function(template <...> void applyFourierFilterFamily)
2340
2341template <class SrcImageIterator, class SrcAccessor,
2342 class FilterType, class DestImage>
2343inline
2347{
2348 applyFourierFilterFamily(src.first, src.second, src.third,
2349 filters, results);
2350}
2351
2352template <class SrcImageIterator, class SrcAccessor,
2353 class FilterType, class DestImage>
2354void applyFourierFilterFamily(SrcImageIterator srcUpperLeft,
2355 SrcImageIterator srcLowerRight, SrcAccessor sa,
2356 const ImageArray<FilterType> &filters,
2357 ImageArray<DestImage> &results)
2358{
2359 int w = int(srcLowerRight.x - srcUpperLeft.x);
2360 int h = int(srcLowerRight.y - srcUpperLeft.y);
2361
2362 FFTWComplexImage workImage(w, h);
2363 copyImage(srcIterRange(srcUpperLeft, srcLowerRight, sa),
2364 destImage(workImage, FFTWWriteRealAccessor<>()));
2365
2366 FFTWComplexImage const & cworkImage = workImage;
2367 applyFourierFilterFamilyImpl(cworkImage.upperLeft(), cworkImage.lowerRight(), cworkImage.accessor(),
2368 filters, results);
2369}
2370
2371template <class FilterType, class DestImage>
2372inline
2374 FFTWComplexImage::const_traverser srcUpperLeft,
2375 FFTWComplexImage::const_traverser srcLowerRight,
2376 FFTWComplexImage::ConstAccessor sa,
2377 const ImageArray<FilterType> &filters,
2378 ImageArray<DestImage> &results)
2379{
2380 int w= srcLowerRight.x - srcUpperLeft.x;
2381
2382 // test for right memory layout (fftw expects a 2*width*height floats array)
2383 if (&(*(srcUpperLeft + Diff2D(w, 0))) == &(*(srcUpperLeft + Diff2D(0, 1))))
2384 applyFourierFilterFamilyImpl(srcUpperLeft, srcLowerRight, sa,
2385 filters, results);
2386 else
2387 {
2388 int h = srcLowerRight.y - srcUpperLeft.y;
2389 FFTWComplexImage workImage(w, h);
2390 copyImage(srcIterRange(srcUpperLeft, srcLowerRight, sa),
2391 destImage(workImage));
2392
2393 FFTWComplexImage const & cworkImage = workImage;
2394 applyFourierFilterFamilyImpl(cworkImage.upperLeft(), cworkImage.lowerRight(), cworkImage.accessor(),
2395 filters, results);
2396 }
2397}
2398
2399template <class FilterType, class DestImage>
2400void applyFourierFilterFamilyImpl(
2401 FFTWComplexImage::const_traverser srcUpperLeft,
2402 FFTWComplexImage::const_traverser srcLowerRight,
2403 FFTWComplexImage::ConstAccessor /*sa*/,
2404 const ImageArray<FilterType> &filters,
2405 ImageArray<DestImage> &results)
2406{
2407 // FIXME: sa is not used
2408 // (maybe check if StandardAccessor, else copy?)
2409
2410 // make sure the filter images have the right dimensions
2411 vigra_precondition((srcLowerRight - srcUpperLeft) == filters.imageSize(),
2412 "applyFourierFilterFamily called with src image size != filters.imageSize()!");
2413
2414 // make sure the result image array has the right dimensions
2415 results.resize(filters.size());
2416 results.resizeImages(filters.imageSize());
2417
2418 // FFT from srcImage to freqImage
2419 int w = int(srcLowerRight.x - srcUpperLeft.x);
2420 int h = int(srcLowerRight.y - srcUpperLeft.y);
2421
2422 FFTWComplexImage freqImage(w, h);
2423 FFTWComplexImage result(w, h);
2424
2425 fftw_plan forwardPlan=
2426 fftw_plan_dft_2d(h, w, (fftw_complex *)&(*srcUpperLeft),
2427 (fftw_complex *)freqImage.begin(),
2428 FFTW_FORWARD, FFTW_ESTIMATE );
2429 fftw_execute(forwardPlan);
2430 fftw_destroy_plan(forwardPlan);
2431
2432 fftw_plan backwardPlan=
2433 fftw_plan_dft_2d(h, w, (fftw_complex *)result.begin(),
2434 (fftw_complex *)result.begin(),
2435 FFTW_BACKWARD, FFTW_ESTIMATE );
2436 typedef typename
2437 NumericTraits<typename DestImage::Accessor::value_type>::isScalar
2438 isScalarResult;
2439
2440 // convolve with filters in freq. domain
2441 for (unsigned int i= 0; i < filters.size(); i++)
2442 {
2443 combineTwoImages(srcImageRange(freqImage), srcImage(filters[i]),
2444 destImage(result), std::multiplies<FFTWComplex<> >());
2445
2446 // FFT back into spatial domain (inplace in destImage)
2447 fftw_execute(backwardPlan);
2448
2449 // normalization (after FFTs), maybe stripping imaginary part
2450 applyFourierFilterImplNormalization(result,
2451 results[i].upperLeft(), results[i].accessor(),
2452 isScalarResult());
2453 }
2454 fftw_destroy_plan(backwardPlan);
2455}
2456
2457/********************************************************/
2458/* */
2459/* fourierTransformReal */
2460/* */
2461/********************************************************/
2462
2463/** \brief Real Fourier transforms for even and odd boundary conditions
2464 (aka. cosine and sine transforms).
2465
2466
2467 If the image is real and has even symmetry, its Fourier transform
2468 is also real and has even symmetry. The Fourier transform of a real image with odd
2469 symmetry is imaginary and has odd symmetry. In either case, only about a quarter
2470 of the pixels need to be stored because the rest can be calculated from the symmetry
2471 properties. This is especially useful, if the original image is implicitly assumed
2472 to have reflective or anti-reflective boundary conditions. Then the "negative"
2473 pixel locations are defined as
2474
2475 \code
2476 even (reflective boundary conditions): f[-x] = f[x] (x = 1,...,N-1)
2477 odd (anti-reflective boundary conditions): f[-1] = 0
2478 f[-x] = -f[x-2] (x = 2,...,N-1)
2479 \endcode
2480
2481 end similar at the other boundary (see the FFTW documentation for details).
2482 This has the advantage that more efficient Fourier transforms that use only
2483 real numbers can be implemented. These are also known as cosine and sine transforms
2484 respectively.
2485
2486 If you use the odd transform it is important to note that in the Fourier domain,
2487 the DC component is always zero and is therefore dropped from the data structure.
2488 This means that index 0 in an odd symmetric Fourier domain image refers to
2489 the <i>first</i> harmonic. This is especially important if an image is first
2490 cosine transformed (even symmetry), then in the Fourier domain multiplied
2491 with an odd symmetric filter (e.g. a first derivative) and finally transformed
2492 back to the spatial domain with a sine transform (odd symmetric). For this to work
2493 properly the image must be shifted left or up by one pixel (depending on whether
2494 the x- or y-axis is odd symmetric) before the inverse transform can be applied.
2495 (see example below).
2496
2497 The real Fourier transform functions are named <tt>fourierTransformReal??</tt>
2498 where the questions marks stand for either <tt>E</tt> or <tt>O</tt> indicating
2499 whether the x- and y-axis is to be transformed using even or odd symmetry.
2500 The same functions can be used for both the forward and inverse transforms,
2501 only the normalization changes. For signal processing, the following
2502 normalization factors are most appropriate:
2503
2504 \code
2505 forward inverse
2506 ------------------------------------------------------------
2507 X even, Y even 1.0 4.0 * (w-1) * (h-1)
2508 X even, Y odd -1.0 -4.0 * (w-1) * (h+1)
2509 X odd, Y even -1.0 -4.0 * (w+1) * (h-1)
2510 X odd, Y odd 1.0 4.0 * (w+1) * (h+1)
2511 \endcode
2512
2513 where <tt>w</tt> and <tt>h</tt> denote the image width and height.
2514
2515 <b> Declarations:</b>
2516
2517 pass 2D array views:
2518 \code
2519 namespace vigra {
2520 template <class SrcTraverser, class SrcAccessor,
2521 class DestTraverser, class DestAccessor>
2522 void
2523 fourierTransformRealEE(SrcTraverser sul, SrcTraverser slr, SrcAccessor src,
2524 DestTraverser dul, DestAccessor dest, fftw_real norm);
2525
2526 fourierTransformRealEO, fourierTransformRealOE, fourierTransformRealOO likewise
2527 }
2528 \endcode
2529
2530 pass \ref ImageIterators and \ref DataAccessors :
2531 \code
2532 namespace vigra {
2533 template <class SrcTraverser, class SrcAccessor,
2534 class DestTraverser, class DestAccessor>
2535 void
2536 fourierTransformRealEE(SrcTraverser sul, SrcTraverser slr, SrcAccessor src,
2537 DestTraverser dul, DestAccessor dest, fftw_real norm);
2538
2539 fourierTransformRealEO, fourierTransformRealOE, fourierTransformRealOO likewise
2540 }
2541 \endcode
2542
2543
2544 use argument objects in conjunction with \ref ArgumentObjectFactories :
2545 \code
2546 namespace vigra {
2547 template <class SrcTraverser, class SrcAccessor,
2548 class DestTraverser, class DestAccessor>
2549 void
2550 fourierTransformRealEE(triple<SrcTraverser, SrcTraverser, SrcAccessor> src,
2551 pair<DestTraverser, DestAccessor> dest, fftw_real norm);
2552
2553 fourierTransformRealEO, fourierTransformRealOE, fourierTransformRealOO likewise
2554 }
2555 \endcode
2556
2557 <b> Usage:</b>
2558
2559 <b>\#include</b> <vigra/fftw3.hxx><br>
2560 Namespace: vigra
2561
2562 \code
2563 vigra::FImage spatial(width,height), fourier(width,height);
2564 ... // fill image with data
2565
2566 // forward cosine transform == reflective boundary conditions
2567 fourierTransformRealEE(srcImageRange(spatial), destImage(fourier), (fftw_real)1.0);
2568
2569 // multiply with a first derivative of Gaussian in x-direction
2570 for(int y = 0; y < height; ++y)
2571 {
2572 for(int x = 1; x < width; ++x)
2573 {
2574 double dx = x * M_PI / (width - 1);
2575 double dy = y * M_PI / (height - 1);
2576 fourier(x-1, y) = fourier(x, y) * dx * std::exp(-(dx*dx + dy*dy) * scale*scale / 2.0);
2577 }
2578 fourier(width-1, y) = 0.0;
2579 }
2580
2581 // inverse transform -- odd symmetry in x-direction, even in y,
2582 // due to symmetry of the filter
2583 fourierTransformRealOE(srcImageRange(fourier), destImage(spatial),
2584 (fftw_real)-4.0 * (width+1) * (height-1));
2585 \endcode
2586*/
2587doxygen_overloaded_function(template <...> void fourierTransformReal)
2588
2589template <class SrcTraverser, class SrcAccessor,
2590 class DestTraverser, class DestAccessor>
2591inline void
2594{
2595 fourierTransformRealEE(src.first, src.second, src.third,
2596 dest.first, dest.second, norm);
2597}
2598
2599template <class SrcTraverser, class SrcAccessor,
2600 class DestTraverser, class DestAccessor>
2601inline void
2602fourierTransformRealEE(SrcTraverser sul, SrcTraverser slr, SrcAccessor src,
2603 DestTraverser dul, DestAccessor dest, fftw_real norm)
2604{
2605 fourierTransformRealWorkImageImpl(sul, slr, src, dul, dest,
2606 norm, FFTW_REDFT00, FFTW_REDFT00);
2607}
2608
2609template <class DestTraverser, class DestAccessor>
2610inline void
2611fourierTransformRealEE(
2612 FFTWRealImage::const_traverser sul,
2613 FFTWRealImage::const_traverser slr,
2614 FFTWRealImage::Accessor src,
2615 DestTraverser dul, DestAccessor dest, fftw_real norm)
2616{
2617 int w = slr.x - sul.x;
2618
2619 // test for right memory layout (fftw expects a width*height fftw_real array)
2620 if (&(*(sul + Diff2D(w, 0))) == &(*(sul + Diff2D(0, 1))))
2621 fourierTransformRealImpl(sul, slr, dul, dest,
2622 norm, FFTW_REDFT00, FFTW_REDFT00);
2623 else
2624 fourierTransformRealWorkImageImpl(sul, slr, src, dul, dest,
2625 norm, FFTW_REDFT00, FFTW_REDFT00);
2626}
2627
2628/********************************************************************/
2629
2630template <class SrcTraverser, class SrcAccessor,
2631 class DestTraverser, class DestAccessor>
2632inline void
2633fourierTransformRealOE(triple<SrcTraverser, SrcTraverser, SrcAccessor> src,
2634 pair<DestTraverser, DestAccessor> dest, fftw_real norm)
2635{
2636 fourierTransformRealOE(src.first, src.second, src.third,
2637 dest.first, dest.second, norm);
2638}
2639
2640template <class SrcTraverser, class SrcAccessor,
2641 class DestTraverser, class DestAccessor>
2642inline void
2643fourierTransformRealOE(SrcTraverser sul, SrcTraverser slr, SrcAccessor src,
2644 DestTraverser dul, DestAccessor dest, fftw_real norm)
2645{
2646 fourierTransformRealWorkImageImpl(sul, slr, src, dul, dest,
2647 norm, FFTW_RODFT00, FFTW_REDFT00);
2648}
2649
2650template <class DestTraverser, class DestAccessor>
2651inline void
2652fourierTransformRealOE(
2653 FFTWRealImage::const_traverser sul,
2654 FFTWRealImage::const_traverser slr,
2655 FFTWRealImage::Accessor src,
2656 DestTraverser dul, DestAccessor dest, fftw_real norm)
2657{
2658 int w = slr.x - sul.x;
2659
2660 // test for right memory layout (fftw expects a width*height fftw_real array)
2661 if (&(*(sul + Diff2D(w, 0))) == &(*(sul + Diff2D(0, 1))))
2662 fourierTransformRealImpl(sul, slr, dul, dest,
2663 norm, FFTW_RODFT00, FFTW_REDFT00);
2664 else
2665 fourierTransformRealWorkImageImpl(sul, slr, src, dul, dest,
2666 norm, FFTW_RODFT00, FFTW_REDFT00);
2667}
2668
2669/********************************************************************/
2670
2671template <class SrcTraverser, class SrcAccessor,
2672 class DestTraverser, class DestAccessor>
2673inline void
2674fourierTransformRealEO(triple<SrcTraverser, SrcTraverser, SrcAccessor> src,
2675 pair<DestTraverser, DestAccessor> dest, fftw_real norm)
2676{
2677 fourierTransformRealEO(src.first, src.second, src.third,
2678 dest.first, dest.second, norm);
2679}
2680
2681template <class SrcTraverser, class SrcAccessor,
2682 class DestTraverser, class DestAccessor>
2683inline void
2684fourierTransformRealEO(SrcTraverser sul, SrcTraverser slr, SrcAccessor src,
2685 DestTraverser dul, DestAccessor dest, fftw_real norm)
2686{
2687 fourierTransformRealWorkImageImpl(sul, slr, src, dul, dest,
2688 norm, FFTW_REDFT00, FFTW_RODFT00);
2689}
2690
2691template <class DestTraverser, class DestAccessor>
2692inline void
2693fourierTransformRealEO(
2694 FFTWRealImage::const_traverser sul,
2695 FFTWRealImage::const_traverser slr,
2696 FFTWRealImage::Accessor src,
2697 DestTraverser dul, DestAccessor dest, fftw_real norm)
2698{
2699 int w = slr.x - sul.x;
2700
2701 // test for right memory layout (fftw expects a width*height fftw_real array)
2702 if (&(*(sul + Diff2D(w, 0))) == &(*(sul + Diff2D(0, 1))))
2703 fourierTransformRealImpl(sul, slr, dul, dest,
2704 norm, FFTW_REDFT00, FFTW_RODFT00);
2705 else
2706 fourierTransformRealWorkImageImpl(sul, slr, src, dul, dest,
2707 norm, FFTW_REDFT00, FFTW_RODFT00);
2708}
2709
2710/********************************************************************/
2711
2712template <class SrcTraverser, class SrcAccessor,
2713 class DestTraverser, class DestAccessor>
2714inline void
2715fourierTransformRealOO(triple<SrcTraverser, SrcTraverser, SrcAccessor> src,
2716 pair<DestTraverser, DestAccessor> dest, fftw_real norm)
2717{
2718 fourierTransformRealOO(src.first, src.second, src.third,
2719 dest.first, dest.second, norm);
2720}
2721
2722template <class SrcTraverser, class SrcAccessor,
2723 class DestTraverser, class DestAccessor>
2724inline void
2725fourierTransformRealOO(SrcTraverser sul, SrcTraverser slr, SrcAccessor src,
2726 DestTraverser dul, DestAccessor dest, fftw_real norm)
2727{
2728 fourierTransformRealWorkImageImpl(sul, slr, src, dul, dest,
2729 norm, FFTW_RODFT00, FFTW_RODFT00);
2730}
2731
2732template <class DestTraverser, class DestAccessor>
2733inline void
2734fourierTransformRealOO(
2735 FFTWRealImage::const_traverser sul,
2736 FFTWRealImage::const_traverser slr,
2737 FFTWRealImage::Accessor src,
2738 DestTraverser dul, DestAccessor dest, fftw_real norm)
2739{
2740 int w = slr.x - sul.x;
2741
2742 // test for right memory layout (fftw expects a width*height fftw_real array)
2743 if (&(*(sul + Diff2D(w, 0))) == &(*(sul + Diff2D(0, 1))))
2744 fourierTransformRealImpl(sul, slr, dul, dest,
2745 norm, FFTW_RODFT00, FFTW_RODFT00);
2746 else
2747 fourierTransformRealWorkImageImpl(sul, slr, src, dul, dest,
2748 norm, FFTW_RODFT00, FFTW_RODFT00);
2749}
2750
2751/*******************************************************************/
2752
2753template <class SrcTraverser, class SrcAccessor,
2754 class DestTraverser, class DestAccessor>
2755void
2756fourierTransformRealWorkImageImpl(SrcTraverser sul, SrcTraverser slr, SrcAccessor src,
2757 DestTraverser dul, DestAccessor dest,
2758 fftw_real norm, fftw_r2r_kind kindx, fftw_r2r_kind kindy)
2759{
2760 FFTWRealImage workImage(slr - sul);
2761 copyImage(srcIterRange(sul, slr, src), destImage(workImage));
2762 FFTWRealImage const & cworkImage = workImage;
2763 fourierTransformRealImpl(cworkImage.upperLeft(), cworkImage.lowerRight(),
2764 dul, dest, norm, kindx, kindy);
2765}
2766
2767
2768template <class DestTraverser, class DestAccessor>
2769void
2770fourierTransformRealImpl(
2771 FFTWRealImage::const_traverser sul,
2772 FFTWRealImage::const_traverser slr,
2773 DestTraverser dul, DestAccessor dest,
2774 fftw_real norm, fftw_r2r_kind kindx, fftw_r2r_kind kindy)
2775{
2776 int w = slr.x - sul.x;
2777 int h = slr.y - sul.y;
2778 BasicImage<fftw_real> res(w, h);
2779
2780 fftw_plan plan = fftw_plan_r2r_2d(h, w,
2781 (fftw_real *)&(*sul), (fftw_real *)res.begin(),
2782 kindy, kindx, FFTW_ESTIMATE);
2783 fftw_execute(plan);
2784 fftw_destroy_plan(plan);
2785
2786 if(norm != 1.0)
2787 transformImage(srcImageRange(res), destIter(dul, dest),
2788 std::bind(std::multiplies<fftw_real>(), 1.0 / norm, std::placeholders::_1));
2789 else
2790 copyImage(srcImageRange(res), destIter(dul, dest));
2791}
2792
2793
2794//@}
2795
2796} // namespace vigra
2797
2798#endif // VIGRA_FFTW3_HXX
Definition basicimage.hxx:265
Two dimensional difference vector.
Definition diff2d.hxx:186
Wrapper class for the FFTW complex types 'fftw_complex'.
Definition fftw3.hxx:132
FFTWReal2Complex< Real >::type complex_type
Definition fftw3.hxx:136
FFTWComplex(fftw_complex const &o)
Definition fftw3.hxx:194
FFTWComplex(value_type const &re=0.0, value_type const &im=0.0)
Definition fftw3.hxx:169
NormType magnitude() const
Definition fftw3.hxx:363
FFTWComplex & operator=(fftwl_complex const &o)
Definition fftw3.hxx:273
FFTWComplex & operator=(FFTWComplex< U > const &o)
Definition fftw3.hxx:246
FFTWComplex(FFTWComplex const &o)
Definition fftw3.hxx:177
value_type NormType
Definition fftw3.hxx:160
FFTWComplex(fftwf_complex const &o)
Definition fftw3.hxx:202
value_type * iterator
Definition fftw3.hxx:152
FFTWComplex & operator=(float o)
Definition fftw3.hxx:291
FFTWComplex operator-() const
Definition fftw3.hxx:353
Real value_type
Definition fftw3.hxx:140
value_type const & const_reference
Definition fftw3.hxx:148
FFTWComplex & operator=(std::complex< T > const &o)
Definition fftw3.hxx:320
FFTWComplex & operator=(FFTWComplex const &o)
Definition fftw3.hxx:236
FFTWComplex(fftwl_complex const &o)
Definition fftw3.hxx:210
value_type const * const_iterator
Definition fftw3.hxx:156
FFTWComplex & operator=(double o)
Definition fftw3.hxx:282
FFTWComplex(std::complex< T > const &o)
Definition fftw3.hxx:219
value_type phase() const
Definition fftw3.hxx:368
FFTWComplex & operator=(fftwf_complex const &o)
Definition fftw3.hxx:264
value_type SquaredNormType
Definition fftw3.hxx:164
FFTWComplex & operator=(long double o)
Definition fftw3.hxx:300
FFTWComplex & operator=(fftw_complex const &o)
Definition fftw3.hxx:255
value_type & reference
Definition fftw3.hxx:144
FFTWComplex(TinyVector< T, 2 > const &o)
Definition fftw3.hxx:228
FFTWComplex(FFTWComplex< U > const &o)
Definition fftw3.hxx:186
FFTWComplex & operator=(TinyVector< T, 2 > const &o)
Definition fftw3.hxx:310
reference operator[](int i)
Definition fftw3.hxx:373
SquaredNormType squaredMagnitude() const
Definition fftw3.hxx:358
const_reference operator[](int i) const
Definition fftw3.hxx:378
int size() const
Definition fftw3.hxx:383
Definition fftw3.hxx:1269
value_type operator()(ITERATOR const &i) const
Read imaginary part at iterator position.
Definition fftw3.hxx:1276
void set(FFTWComplex< R > const &v, ITERATOR const &i, DIFFERENCE d) const
Write imaginary part at offset from iterator position into a scalar.
Definition fftw3.hxx:1306
void set(value_type const &v, ITERATOR const &i, DIFFERENCE d) const
Write imaginary part at offset from iterator position from a scalar.
Definition fftw3.hxx:1294
void set(value_type const &v, ITERATOR const &i) const
Write imaginary part at iterator position from a scalar.
Definition fftw3.hxx:1288
Real value_type
The accessor's value type.
Definition fftw3.hxx:1272
void set(FFTWComplex< R > const &v, ITERATOR const &i) const
Write imaginary part at iterator position into a scalar.
Definition fftw3.hxx:1300
value_type operator()(ITERATOR const &i, DIFFERENCE d) const
Read imaginary part at offset from iterator position.
Definition fftw3.hxx:1282
Definition fftw3.hxx:1404
value_type operator()(ITERATOR const &i) const
Read natural log of magnitude at iterator position.
Definition fftw3.hxx:1411
Real value_type
The accessor's value type.
Definition fftw3.hxx:1407
value_type operator()(ITERATOR const &i, DIFFERENCE d) const
Read natural log of magnitude at offset from iterator position.
Definition fftw3.hxx:1417
Definition fftw3.hxx:1378
value_type operator()(ITERATOR const &i) const
Read magnitude at iterator position.
Definition fftw3.hxx:1385
Real value_type
The accessor's value type.
Definition fftw3.hxx:1381
value_type operator()(ITERATOR const &i, DIFFERENCE d) const
Read magnitude at offset from iterator position.
Definition fftw3.hxx:1391
Definition fftw3.hxx:1430
value_type operator()(ITERATOR const &i) const
Read phase at iterator position.
Definition fftw3.hxx:1437
Real value_type
The accessor's value type.
Definition fftw3.hxx:1433
value_type operator()(ITERATOR const &i, DIFFERENCE d) const
Read phase at offset from iterator position.
Definition fftw3.hxx:1443
Definition fftw3.hxx:1218
value_type operator()(ITERATOR const &i) const
Read real part at iterator position.
Definition fftw3.hxx:1226
void set(FFTWComplex< R > const &v, ITERATOR const &i, DIFFERENCE d) const
Write real part at offset from iterator position into a scalar.
Definition fftw3.hxx:1256
void set(value_type const &v, ITERATOR const &i, DIFFERENCE d) const
Write real part at offset from iterator position from a scalar.
Definition fftw3.hxx:1244
void set(value_type const &v, ITERATOR const &i) const
Write real part at iterator position from a scalar.
Definition fftw3.hxx:1238
Real value_type
The accessor's value type.
Definition fftw3.hxx:1222
void set(FFTWComplex< R > const &v, ITERATOR const &i) const
Write real part at iterator position into a scalar.
Definition fftw3.hxx:1250
value_type operator()(ITERATOR const &i, DIFFERENCE d) const
Read real part at offset from iterator position.
Definition fftw3.hxx:1232
Definition fftw3.hxx:1352
value_type operator()(ITERATOR const &i) const
Read squared magnitude at iterator position.
Definition fftw3.hxx:1359
Real value_type
The accessor's value type.
Definition fftw3.hxx:1355
value_type operator()(ITERATOR const &i, DIFFERENCE d) const
Read squared magnitude at offset from iterator position.
Definition fftw3.hxx:1365
Definition fftw3.hxx:1321
void set(value_type const &v, ITERATOR const &i, DIFFERENCE d) const
Definition fftw3.hxx:1339
void set(value_type const &v, ITERATOR const &i) const
Definition fftw3.hxx:1330
Real value_type
The accessor's value type.
Definition fftw3.hxx:1324
Class for a single RGB value.
Definition rgbvalue.hxx:128
NormType magnitude() const
Definition rgbvalue.hxx:307
SquaredNormType squaredMagnitude() const
Definition rgbvalue.hxx:313
void transformImage(...)
Apply unary point transformation to each pixel.
void fourierTransform(...)
Compute forward and inverse Fourier transforms.
void fourierTransformInverse(...)
Compute inverse Fourier transforms.
void applyFourierFilter(...)
Apply a filter (defined in the frequency domain) to an image.
void applyFourierFilterFamily(...)
Apply an array of filters (defined in the frequency domain) to an image.
Diff2D operator+(Diff2D const &a, Diff2D const &b)
Definition diff2d.hxx:725
R arg(const FFTWComplex< R > &a)
phase
Definition fftw3.hxx:1009
void moveDCToUpperLeft(...)
Rearrange the quadrants of a Fourier image so that the origin is in the image's upper left.
R imag(const FFTWComplex< R > &a)
imaginary part
Definition fftw3.hxx:1023
FFTWComplex< R >::NormType norm(const FFTWComplex< R > &a)
norm (= magnitude)
Definition fftw3.hxx:1037
BasicImage< FFTWComplex<> > FFTWComplexImage
Definition fftw3.hxx:1192
FFTWComplex< R > conj(const FFTWComplex< R > &a)
complex conjugate
Definition fftw3.hxx:1030
R real(const FFTWComplex< R > &a)
real part
Definition fftw3.hxx:1016
FFTWComplex< R > & operator+=(FFTWComplex< R > &a, const FFTWComplex< R > &b)
add-assignment
Definition fftw3.hxx:859
FFTWComplex< R > & operator-=(FFTWComplex< R > &a, const FFTWComplex< R > &b)
subtract-assignment
Definition fftw3.hxx:867
void combineTwoImages(...)
Combine two source images into destination image.
T sign(T t)
The sign function.
Definition mathutil.hxx:591
FFTWComplex< R > & operator*=(FFTWComplex< R > &a, const FFTWComplex< R > &b)
multiply-assignment
Definition fftw3.hxx:875
FFTWComplex< R > & operator/=(FFTWComplex< R > &a, const FFTWComplex< R > &b)
divide-assignment
Definition fftw3.hxx:884
void fourierTransformReal(...)
Real Fourier transforms for even and odd boundary conditions (aka. cosine and sine transforms).
FFTWComplex< R >::SquaredNormType squaredNorm(const FFTWComplex< R > &a)
squared norm (= squared magnitude)
Definition fftw3.hxx:1044
void moveDCToCenter(...)
Rearrange the quadrants of a Fourier image so that the origin is in the image center.
bool operator==(FFTWComplex< R > const &a, const FFTWComplex< R > &b)
equal
Definition fftw3.hxx:825
LinearIntensityTransform< DestValueType, Multiplier > linearIntensityTransform(Multiplier scale, DestValueType offset)
Apply a linear transform to the source pixel values.
Definition transformimage.hxx:800
BasicImage< fftw_real > FFTWRealImage
Definition fftw3.hxx:1132
bool operator!=(FFTWComplex< R > const &a, const FFTWComplex< R > &b)
not equal
Definition fftw3.hxx:841
Diff2D operator-(Diff2D const &a, Diff2D const &b)
Definition diff2d.hxx:697
void copyImage(...)
Copy source image into destination image.
Export associated information for each image iterator.
Definition iteratortraits.hxx:110

© Ullrich Köthe (ullrich.koethe@iwr.uni-heidelberg.de)
Heidelberg Collaboratory for Image Processing, University of Heidelberg, Germany

html generated using doxygen and Python
vigra 1.11.2