学习opencv 第六章 习题十三

用傅里叶变换加速卷积,直接上代码，Mat版是Copy他人的。

CvMat版

#include "stdafx.h"
#include "cv.h"
#include "highgui.h"
#include <iostream>

using namespace cv;
using namespace std;

void speedy_convolution(const CvMat* A,const CvMat* B,CvMat* C);

int main()
{
    IplImage* img=cvLoadImage("C:/Users/shark/Desktop/fruits.jpg",0);
    CvMat* src=cvCreateMat(img->height,img->width,CV_32FC1);
    /*int data;
    for(int i=0;i<img->height;i++)
    {
        for(int j=0;j<img->width;j++)
        {
            data=img->imageData[i*img->widthStep+j];
            cvmSet(src,i,j,data);
        }
    }*/
    //必须归一化矩阵的值为0-1之间(缩放比例在1/255.0附近效果最好,太小最后会全黑,接近1或大于1几乎是全白;
    //(还未深入了解函数cvConvertScale的机理),缩放比例不能为1,打出目标图像的像素有正有负
    cvConvertScale(img,src,1/255.0,0);  
    

    CvMat* kernel=cvCreateMat(3,3,CV_32FC1);
    cvSetReal2D(kernel,0,0,1.0/16);    cvSetReal2D(kernel,0,1,2.0/16); cvSetReal2D(kernel,0,2,1.0/16);    //注意设置值时必须加个.0否则1/16的值0
    cvSetReal2D(kernel,1,0,2.0/16);    cvSetReal2D(kernel,1,1,4.0/16); cvSetReal2D(kernel,1,2,2.0/16);
    cvSetReal2D(kernel,2,0,1.0/16);    cvSetReal2D(kernel,2,1,2.0/16); cvSetReal2D(kernel,2,2,1.0/16);
    CvMat* C=cvCreateMat((src->rows+kernel->rows-1),(src->cols+kernel->cols-1),src->type);
    speedy_convolution(src,kernel,C);

    IplImage* img_src=cvCreateImage(cvGetSize(src),IPL_DEPTH_32F,1);
    cvGetImage(src,img_src);
    IplImage* img_dst=cvCreateImage(cvGetSize(C),IPL_DEPTH_32F,1);
    cvGetImage(C,img_dst);

    cvNamedWindow("img_src");
    cvShowImage("img_src",img_src);
    cvNamedWindow("img");
    cvShowImage("img",img);
    cvNamedWindow("dst");
    cvShowImage("dst",img_dst);
    cvWaitKey();
    return 0;
}

void speedy_convolution(
    const CvMat* A,
    const CvMat* B,
    CvMat* C
    ){
        int dft_M=cvGetOptimalDFTSize(A->rows+B->rows-1);
        int dft_N=cvGetOptimalDFTSize(A->cols+B->cols-1);

        CvMat *dft_A=cvCreateMat(dft_M,dft_N,A->type);
        CvMat *dft_B=cvCreateMat(dft_M,dft_N,B->type);
        CvMat tmp;
        cvGetSubRect(dft_A,&tmp,cvRect(0,0,A->cols,A->rows));
        cvCopy(A,&tmp);
        cvGetSubRect(dft_A,&tmp,cvRect(A->cols,0,dft_A->cols-A->cols,A->rows));
        cvZero(&tmp);
        cvDFT(dft_A,dft_A,CV_DXT_FORWARD,A->rows);

        cvGetSubRect(dft_B,&tmp,cvRect(0,0,B->cols,B->rows));
        cvCopy(B,&tmp);
        cvGetSubRect(dft_B,&tmp,cvRect(B->cols,0,dft_B->cols-B->cols,B->rows));
        cvZero(&tmp);
        cvDFT(dft_B,dft_B,CV_DXT_FORWARD,B->rows);

        cvMulSpectrums(dft_A,dft_B,dft_A,0);

        cvDFT(dft_A,dft_A,CV_DXT_INV_SCALE,C->rows);
        cvGetSubRect(dft_A,&tmp,cvRect(0,0,C->cols,C->rows));
        cvCopy(&tmp,C);
        cvReleaseMat(&dft_A);
        cvReleaseMat(&dft_B);
}

Mat版

#include "opencv2/core/core.hpp"
#include "opencv2/imgproc/imgproc.hpp"
#include "opencv2/highgui/highgui.hpp"
#include <iostream>

using namespace cv;
using namespace std;

//http://docs.opencv.org/modules/core/doc/operations_on_arrays.html#dft[2]
void convolveDFT(Mat A, Mat B, Mat& C)
{
    // reallocate the output array if needed
    C.create(abs(A.rows - B.rows)+1, abs(A.cols - B.cols)+1, A.type());
    Size dftSize;
    // calculate the size of DFT transform
    dftSize.width = getOptimalDFTSize(A.cols + B.cols - 1);
    dftSize.height = getOptimalDFTSize(A.rows + B.rows - 1);

    // allocate temporary buffers and initialize them with 0's
    Mat tempA(dftSize, A.type(), Scalar::all(0));
    Mat tempB(dftSize, B.type(), Scalar::all(0));

    // copy A and B to the top-left corners of tempA and tempB, respectively
    Mat roiA(tempA, Rect(0,0,A.cols,A.rows));
    A.copyTo(roiA);
    Mat roiB(tempB, Rect(0,0,B.cols,B.rows));
    B.copyTo(roiB);

    // now transform the padded A & B in-place;
    // use "nonzeroRows" hint for faster processing
    dft(tempA, tempA, 0, A.rows);
    dft(tempB, tempB, 0, B.rows);

    // multiply the spectrums;
    // the function handles packed spectrum representations well
    mulSpectrums(tempA, tempB, tempA, DFT_COMPLEX_OUTPUT);
    //mulSpectrums(tempA, tempB, tempA, DFT_REAL_OUTPUT);

    // transform the product back from the frequency domain.
    // Even though all the result rows will be non-zero,
    // you need only the first C.rows of them, and thus you
    // pass nonzeroRows == C.rows
    dft(tempA, tempA, DFT_INVERSE + DFT_SCALE, C.rows);

    // now copy the result back to C.
    tempA(Rect(0, 0, C.cols, C.rows)).copyTo(C);

    // all the temporary buffers will be deallocated automatically
}


int main(int argc, char* argv[])
{
    const char* filename = argc >=2 ? argv[1] : "Lenna.png";

    Mat I = imread(filename, CV_LOAD_IMAGE_GRAYSCALE);
    if( I.empty())
        return -1;

    Mat kernel = (Mat_<float>(3,3) << 1, 1, 1, 1, 1, 1, 1, 1, 1);
    cout << kernel;

    Mat floatI = Mat_<float>(I);// change image type into float
    Mat filteredI;
    convolveDFT(floatI, kernel, filteredI);
    
    normalize(filteredI, filteredI, 0, 1, CV_MINMAX); // Transform the matrix with float values into a
                                            // viewable image form (float between values 0 and 1).
    imshow("image", I);
    imshow("filtered", filteredI);
    waitKey(0);

}

//一是输出Mat C应声明为引用；二是其中的mulSpectrums函数的第四个参数flag值没有指定，应指定为DFT_COMPLEX_OUTPUT或是DFT_REAL_OUTPUT.

//main函数中首先按灰度图读入图像，然后创造一个平滑核kernel，将输入图像转换成float类型(注意这步是必须的，因为dft只能处理浮点数)，在调用convolveDFT求出卷积结果后，将卷积结果归一化方便显示观看。

//需要注意的是，一般求法中，利用核游走整个图像进行卷积运算，实际上进行的是相关运算，真正意义上的卷积，应该首先把核翻转180度，再在整个图像上进行游走。OpenCV中的filter2D实际上做的也只是相关，而非卷积。