svm颜色分类

/* 这个是完整版本，添加标签，分类 */ #include <opencv2/core/core.hpp> #include <opencv2/highgui/highgui.hpp> #include <opencv2/ml/ml.hpp> #include<iostream> #include<opencv2\opencv.hpp> #include <windows.h> using namespace std; using namespace cv; using namespace cv::ml; void on_mouse(int event, int x, int y, int flags, void* ustc); //鼠标取样本点 void scale_Sampling(const Mat &src, Mat &dst, double Ratio); int main(int argc, char *argv[]) { long red_numb = 0, yellow_numb = 0, green_numb = 0, blue_numb = 0, back_numb = 0; //------------------------坐标信息----------------------// //红色：x245,y211 //绿色：x500,y710 //黄色：x770,y60 //蓝色：x1040,y490 //背景：x13,y246 int redx = 245, redy = 211; int greenx = 500, greeny = 710; int yellowx = 770, yellowy = 60; int bluex = 1040, bluey = 490; int backx = 13, backy = 150; int radi = 11; Mat arry[10]; String a = "C:/Users/ncutl/Desktop/siasuna.png"; Mat ori_image = imread(a); Point c(0, 0), b(128, 100); Rect rect1(c, b); Mat roi1; ori_image(rect1).copyTo(roi1); // copy the region rect1 from the image to roi1 imshow("1", roi1); //namedWindow("ori", CV_WINDOW_NORMAL); //imshow("ori", ori_image); //setMouseCallback("ori", on_mouse, 0); /*Mat Red_Mat(50, 50, CV_8UC3, Scalar(0, 0, 255)); cout << endl << Red_Mat << endl << endl; uchar *p = Red_Mat.data; //waitKey(0); Mat T3_M(Red_Mat.rows*Red_Mat.cols, 3, CV_8UC1, p);//转化成一维 cout << "T3_M: " <<endl<< T3_M << endl; */ //--------------------红色训练集---------------------// //rectangle(red, cv::Rect(0, 0, 200, 150), cv::Scalar(255, 0, 0), 10); Mat red_roi_uf(ori_image, Rect(redx, redy, 100, 100)); Mat red_roi; DWORD tim1 = GetTickCount(); medianBlur(red_roi_uf, red_roi, radi); DWORD tim2 = GetTickCount(); cout << tim2 - tim1 << endl; //cout << red_roi << endl << endl; Mat red_roi_convert; red_roi.convertTo(red_roi_convert, CV_32FC1); Mat red_roi_data(red_roi_convert.rows*red_roi_convert.cols, 3, CV_32FC1, red_roi_convert.data); //cout << red_roi_data << endl << endl; Mat red_label = Mat(red_roi_convert.rows*red_roi_convert.cols, 1, CV_32SC1, Scalar::all(1)); //cout << red_label << endl << endl; imshow("红色", red_roi); //-------------------黄色训练集-----------------// Mat yellow_roi_uf(ori_image, Rect(yellowx, yellowy, 100, 100)); Mat yellow_roi; tim1 = GetTickCount(); medianBlur(yellow_roi_uf, yellow_roi, radi); tim2 = GetTickCount(); cout << tim2 - tim1 << endl; //cout << red_roi << endl << endl; Mat yellow_roi_convert; yellow_roi.convertTo(yellow_roi_convert, CV_32FC1); Mat yellow_roi_data(yellow_roi_convert.rows*yellow_roi_convert.cols, 3, CV_32FC1, yellow_roi_convert.data); //cout << yellow_roi_data << endl << endl; Mat yellow_label = Mat(yellow_roi_convert.rows*yellow_roi_convert.cols, 1, CV_32SC1, Scalar::all(2)); imshow("黄色", yellow_roi); //------------------------绿色训练集---------------// Mat green_roi_uf(ori_image, Rect(greenx, greeny, 100, 100)); Mat green_roi; tim1 = GetTickCount(); medianBlur(green_roi_uf, green_roi, radi); tim2 = GetTickCount(); cout << tim2 - tim1 << endl; //cout << green_roi << endl << endl; Mat green_roi_convert; green_roi.convertTo(green_roi_convert, CV_32FC1); Mat green_roi_data(green_roi_convert.rows*green_roi_convert.cols, 3, CV_32FC1, green_roi_convert.data); //cout << green_roi_data << endl << endl; Mat green_label = Mat(green_roi_convert.rows*green_roi_convert.cols, 1, CV_32SC1, Scalar::all(3)); imshow("绿色", green_roi); //cout << green_label << endl; //-----------------------蓝色训练集---------------// Mat blue_roi_uf(ori_image, Rect(bluex, bluey, 100, 100)); Mat blue_roi; tim1 = GetTickCount(); medianBlur(blue_roi_uf, blue_roi, radi); tim2 = GetTickCount(); cout << tim2 - tim1 << endl; //cout << blue_roi << endl << endl; Mat blue_roi_convert; blue_roi.convertTo(blue_roi_convert, CV_32FC1); Mat blue_roi_data(blue_roi_convert.rows*blue_roi_convert.cols, 3, CV_32FC1, blue_roi_convert.data); //cout << blue_roi_data << endl << endl; Mat blue_label = Mat(blue_roi_convert.rows*blue_roi_convert.cols, 1, CV_32SC1, Scalar::all(4)); imshow("蓝色", blue_roi); //---------------------背景训练集-----------------// Mat back_roi_uf(ori_image, Rect(backx, backy, 80, 180)); Mat back_roi; tim1 = GetTickCount(); medianBlur(back_roi_uf, back_roi, radi); tim2 = GetTickCount(); cout << tim2 - tim1 << endl; //cout << back_roi << endl << endl; Mat back_roi_convert; back_roi.convertTo(back_roi_convert, CV_32FC1); Mat back_roi_data(back_roi_convert.rows*back_roi_convert.cols, 3, CV_32FC1, back_roi_convert.data); //cout << back_roi_data << endl << endl; Mat back_label = Mat(back_roi_convert.rows*back_roi_convert.cols, 1, CV_32SC1, Scalar::all(5)); imshow("背景", back_roi); //setMouseCallback("红色", on_mouse, 0); //-----------------合并所有的样本点，作为训练数据----------------------// Mat train_data, train_label; vconcat(red_roi_data, yellow_roi_data, train_data); vconcat(train_data, green_roi_data, train_data); vconcat(train_data, blue_roi_data, train_data); vconcat(train_data, back_roi_data, train_data); vconcat(red_label, yellow_label, train_label); vconcat(train_label, green_label, train_label); vconcat(train_label, blue_label, train_label); vconcat(train_label, back_label, train_label); //cout << train_data << endl; //cout << train_label << endl; // 设置参数 Ptr<SVM> svm = SVM::create(); svm->setType(SVM::C_SVC); svm->setKernel(SVM::POLY); //svm->setNu(0.5); svm->setGamma(100); //svm->setC(100); svm->setDegree(0.1); svm->setTermCriteria(TermCriteria(TermCriteria::MAX_ITER, 100, 1e-6)); // 训练分类器 Ptr<TrainData> tData = TrainData::create(train_data, ROW_SAMPLE, train_label); svm->train(tData); cout << "训练完成" << endl << endl; Vec3b green(0, 255, 0), blue(255, 0, 0), red(0, 0, 255), yellow(0, 255, 255), black(0, 0, 0); Mat test_img_uf = imread("C:\\Users\\ncutl\\Desktop\\22.png"); Mat test_img; Mat samp_ed; scale_Sampling(test_img_uf, test_img, 0.3); //imwrite("C:\\Users\\ncutl\\Desktop\\22_samp.png", test_img); DWORD star_time = GetTickCount(); tim1 = GetTickCount(); //transpose(test_img_uf, test_img); medianBlur(test_img, test_img, 3); //boxFilter(test_img_uf, test_img, -1, Size(5, 5)); tim2 = GetTickCount(); cout << tim2 - tim1 << endl; /*Mat copyim(ori_image_uf, Rect(244, 208, 140, 140)); Mat test_img = copyim.clone();*/ //pixel.val[0]; //cout << (int)green.val[0] << endl; //CvScalar s = cvGet2D(&red, 200, 201); //cout << '\t' << s.val[1] << '\t' << s.val[2] << endl;//输出B值，G值，R值 //Mat sampleMat = (Mat_<float>(1, 3) << temp.at<Vec3b>(j, i)[0], temp.at<Vec3b>(j, i)[1], temp.at<Vec3b>(j, i)[2]); //----------------单个像素测试----------------// for (int i = 0; i < test_img.rows; i++) for (int j = 0; j < test_img.cols; j++) { Vec3b pixel = test_img.at<Vec3b>(i, j); float a_t = pixel[0]; float b_t = pixel[1]; float c_t = pixel[2]; //cout << a_t << endl << b_t << endl << c_t << endl; Mat sampleMat = (Mat_<float>(1, 3) << a_t, b_t, c_t); int response = svm->predict(sampleMat); if (response == 1) { test_img.at<Vec3b>(i, j) = red; red_numb++; } if (response == 2) { test_img.at<Vec3b>(i, j) = yellow; yellow_numb++; } if (response == 3) { test_img.at<Vec3b>(i, j) = green; green_numb++; } if (response == 4) { test_img.at<Vec3b>(i, j) = blue; blue_numb++; } if (response == 5) //cout << "beijing " << endl; { test_img.at<Vec3b>(i, j) = black; back_numb++; }