图像处理之基础---卷积及其快速算法的C++实现

时间:2021-02-21 00:17:51

头文件:

/*
* Copyright (c) 2008-2011 Zhang Ming (M. Zhang), zmjerry@163.com
*
* This program is free software; you can redistribute it and/or modify it
* under the terms of the GNU General Public License as published by the
* Free Software Foundation, either version 2 or any later version.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are met:
*
* 1. Redistributions of source code must retain the above copyright notice,
* this list of conditions and the following disclaimer.
*
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* This program is distributed in the hope that it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
* more details. A copy of the GNU General Public License is available at:
* http://www.fsf.org/licensing/licenses
*/ /*****************************************************************************
* convolution.h
*
* Linear convolution and polynomial multiplication.
*
* The convolution routine "conv" is implemented by it's definition in time
* domain. If the sequence to be convoluted are long, you should use the
* fast convolution algorithm "fastConv", which is implemented in frequency
* domain by usin FFT.
*
* Zhang Ming, 2010-01, Xi'an Jiaotong University.
*****************************************************************************/ #ifndef CONVOLUTION_H
#define CONVOLUTION_H #include <vector.h>
#include <fft.h>
#include <utilities.h> namespace splab
{ template<typename Type> Vector<Type> conv( const Vector<Type>&,
const Vector<Type>& );
template<typename Type> Vector<Type> convolution( const Vector<Type>&,
const Vector<Type>& ); template<typename Type> Vector<Type> fastConv( const Vector<Type>&,
const Vector<Type>& ); #include <convolution-impl.h> }
// namespace splab #endif
// CONVOLUTION_H

实现文件:

/*
* Copyright (c) 2008-2011 Zhang Ming (M. Zhang), zmjerry@163.com
*
* This program is free software; you can redistribute it and/or modify it
* under the terms of the GNU General Public License as published by the
* Free Software Foundation, either version 2 or any later version.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are met:
*
* 1. Redistributions of source code must retain the above copyright notice,
* this list of conditions and the following disclaimer.
*
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* This program is distributed in the hope that it will be useful, but WITHOUT
* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
* more details. A copy of the GNU General Public License is available at:
* http://www.fsf.org/licensing/licenses
*/ /*****************************************************************************
* convolution-impl.h
*
* Implementation for linear convolution.
*
* Zhang Ming, 2010-01, Xi'an Jiaotong University.
*****************************************************************************/ /**
* convolution and ploynonal multiplication.
*/
template <typename Type>
Vector<Type> conv( const Vector<Type> &signal, const Vector<Type> &filter )
{
if( signal.dim() < filter.dim() )
return convolution( filter, signal );
else
return convolution( signal, filter );
} template <typename Type>
Vector<Type> convolution( const Vector<Type> &signal, const Vector<Type> &filter )
{
int sigLength = signal.dim();
int filLength = filter.dim();
assert( sigLength >= filLength ); int length = sigLength + filLength - 1;
Vector<Type> x(length); for( int i=1; i<=length; ++i )
{
x(i) = 0;
if( i < filLength )
for( int j=1; j<=i; ++j )
x(i) += filter(j) * signal(i-j+1);
else if( i <= sigLength )
for( int j=1; j<=filLength; ++j )
x(i) += filter(j) * signal(i-j+1);
else
for( int j=i-sigLength+1; j<=filLength; ++j )
x(i) += filter(j) * signal(i-j+1);
}
return x;
} /**
* Fast convolution by FFT.
*/
template<typename Type>
Vector<Type> fastConv( const Vector<Type> &xn, const Vector<Type> &yn )
{
int M = xn.dim(),
N = yn.dim(); Vector<Type> xnPadded = wextend( xn, N-1, "right", "zpd" ),
ynPadded = wextend( yn, M-1, "right", "zpd" );
return ifftc2r( fft(xnPadded) * fft(ynPadded) ); // Vector< complex<Type> > Zk = fft(xnPadded) * fft(ynPadded);
// return ifftc2r(Zk); // return ifftc2r( fft(wextend(xn,N-1,"right","zpd")) * fft(wextend(yn,M-1,"right","zpd")) );
}

测试代码:

/*****************************************************************************
* convolution.cpp
*
* Convolution testing.
*
* Zhang Ming, 2010-01, Xi'an Jiaotong University.
*****************************************************************************/ #define BOUNDS_CHECK #include <iostream>
#include <convolution.h> using namespace std;
using namespace splab; typedef double Type;
const int M = 3;
const int N = 5; int main()
{
Vector<Type> xn( M ), yn( N );
Vector<Type> zn; for( int i=0; i<M; ++i )
xn[i] = i;
for( int i=0; i<N; ++i )
yn[i] = i-N/2; // convolution
zn = conv( xn, yn );
cout << "xn: " << xn << endl << "yn: " << yn << endl;
cout << "convolution of xn and yn: " << zn << endl;
zn = fastConv( xn, yn );
cout << "fast convolution of xn and yn: " << zn << endl; return 0;
}

运行结果:

xn:  size: 3 by 1
0
1
2 yn: size: 5 by 1
-2
-1
0
1
2 convolution of xn and yn: size: 7 by 1
0
-2
-5
-2
1
4
4 fast convolution of xn and yn: size: 7 by 1
-2.53765e-016
-2
-5
-2
1
4
4 Process returned 0 (0x0) execution time : 0.078 s
Press any key to continue. http://my.oschina.net/zmjerry/blog/3671http://v.youku.com/v_show/id_XMTMwNDI1NDAw.html