数字信号实验：使用C++与OpenCV实现DCT变换压缩图片

# include<stdio.h> # include<math.h> # include<stdlib.h> # include<time.h> #include <opencv2/core/core.hpp> #include <opencv2/highgui/highgui.hpp> #include <iostream> # define N 8 # define PI 3.1415926 using namespace std; using namespace cv; void dct2(Mat& A, Mat& AT, Mat& image, Mat&  );              //dct正变换  void f_dct2(Mat& A, Mat& AT, Mat& image);                   //dct反变换 Mat TMatrix(Mat, int);                                      //转置矩阵 Mat dctmtx(int n);                                          //返回N×N的DCT变换矩阵 Mat compression(int,int,Mat image);                         //图像单通道压缩函数 int main() {     int i, j;     vector<Mat>channels;     //Opencv读入图片，分别用于压缩和比较     Mat Image = imread("bottle477.jpg");     Mat ori_Image = imread("bottle477.jpg");     Image.convertTo(Image, CV_64F);     //分离彩色通道     split(Image, channels);     Mat imageBlueChannel = channels.at(0);     Mat imageGreenChannel = channels.at(1);     Mat imageRedChannel = channels.at(2);     merge(channels, Image);     //R,G,B三通道分别进行压缩     Mat B = compression(Image.cols, Image.rows,imageBlueChannel);     Mat G = compression(Image.cols, Image.rows,imageGreenChannel);     Mat R = compression(Image.cols, Image.rows,imageRedChannel);     //三通道合并     channels.at(0) = Mat_<uchar>(B);     channels.at(1) = Mat_<uchar>(G);     channels.at(2) = Mat_<uchar>(R);     merge(channels, Image);     imshow("DstImage",Image);     imshow("OriImage", ori_Image);     waitKey(0);     return 0; } Mat compression(int r, int c, Mat image) {     Mat A = dctmtx(N);     Mat AT = TMatrix(A, N);     //Mask定义     Mat Mask = (Mat_<double>(N, N)         <<  1, 1, 1, 0, 0, 0, 0, 0,              1, 1, 0, 0, 0, 0, 0, 0,              1, 0, 0, 0, 0, 0, 0, 0,              0, 0, 0, 0, 0, 0, 0, 0,              0, 0, 0, 0, 0, 0, 0, 0,              0, 0, 0, 0, 0, 0, 0, 0,              0, 0, 0, 0, 0, 0, 0, 0,              0, 0, 0, 0, 0, 0, 0, 0);         //DCT变换和Mask矩阵量化     dct2(A, AT, image, Mask);     //DCT逆变换     f_dct2(A, AT, image);     return image; } //生成转换矩阵 Mat dctmtx(int num) {     int i, j;     //初始化矩阵，定义矩阵     Mat A = Mat::zeros(N, N, CV_64F);     //按照公式，生成正交矩阵A     double x0 = sqrt(1.0 / num);     double x1 = sqrt(2.0 / num);     for (i = 0; i < N; ++i) {         for (j = 0; j < N; ++j) {              if (i == 0) {                  A.at<double>(i,j) =  x0 * cos((2 * j + 1) * PI * i / (2 * N));              }              else {                  A.at<double>(i, j) = x1 * cos((2 * j + 1) * PI * i / (2 * N));              }         }     }     return A; } //矩阵转置 Mat TMatrix(Mat A, int n) {     int i, j;     Mat tmp;     tmp = Mat::zeros(A.size(), A.type());     for (j = 0; j < n; ++j)         for (i = 0; i < n; ++i)              tmp.at<double>(i, j) = A.at<double>(j, i);     return tmp; } //DCT变换 void dct2(Mat& A,Mat& AT, Mat& image, Mat& mask) {     for (int i = 0; i < (image.rows - image.rows % N); i += N) {         for (int j = 0; j < (image.cols - image.cols % N); j += N) {              image(Range(i, i + N), Range(j, j + N)) = A * image(Range(i, i + N), Range(j, j + N)) * AT;              image(Range(i, i + N), Range(j, j + N)) = mask.mul(image(Range(i, i + N), Range(j, j + N)));         }     } } //DCT逆变换 void f_dct2(Mat& A, Mat& AT, Mat& image) {     for (int i = 0; i < (image.rows - image.rows % N); i += N) {         for (int j = 0; j < (image.cols - image.cols % N); j += N) {              image(Range(i, i + N), Range(j, j + N)) = AT * image(Range(i, i + N), Range(j, j + N)) * A;         }     } }