java svd_SVD分解 opencv实现(示例代码)

头文件

#ifndef DEBUG_LRN_SVD_H

#define DEBUG_LRN_SVD_H

#include

#include

#include

#include

#include

using namespace std;

using namespace cv;

int test_SVD();

void print_matrix(const vector>& vec);

#endif //DEBUG_LRN_SVD_H

源文件

#include "svd.h"

void print_matrix(const vector>& vec){

if(vec.empty()){

return;

}

for ( auto row: vec){

if(row.empty()){

return;

}

for ( auto elem: row){

cout << elem << ", ";

}

cout << endl;

}

cout << endl;

}

int test_SVD()

{

std::vector<:vector>> vec{ { 0.68f, 0.597f },

{ -0.211f, 0.823f },

{ 0.566f, -0.605f } };

cout << "source matrix：" << endl;

print_matrix(vec);

// cout << "C++ implement SVD:" << endl;

// vector> matD, matU, matVt;

// if (svd(vec, matD, matU, matVt) != 0) {

// cout << "C++ implement singular value decomposition fail!" << endl;

// return -1;

// }

// cout << "singular values:" << endl;

// print_matrix(matD);

// cout << "left singular vectors:" << endl;

// print_matrix(matU);

// cout << "transposed matrix of right singular values:" << endl;

// print_matrix(matVt);

cout << "opencv singular value decomposition:" << endl;

const auto rows = (int)vec.size();

const auto cols = (int)vec[0].size();

cv::Mat mat(rows, cols, CV_32FC1);

for (int y = 0; y < rows; ++y) {

for (int x = 0; x < cols; ++x) {

mat.at(y, x) = vec.at(y).at(x);

}

}

cv::Mat matD, matU, matVt;

cv::SVD::compute(mat, matD, matU, matVt, 4);

cout << "singular values:" << endl;

cout << matD << endl;

cout << "left singular vectors:" << endl;

cout << matU << endl;

cout << "transposed matrix of right singular values:" << endl;

cout << matVt << endl;

cv::Mat matUt, matV;

cv::transpose(matU, matUt);

cv::transpose(matVt, matV);

cout << "verify if the result is correct:" << endl;

cv::Mat reconMatD = matUt * mat * matV;

cout << reconMatD << endl;

return 0;

}