C# OPENCV几何形状识别

【实例截图】

【核心代码】

//opencv2.4.13
#include "opencv2/core/core.hpp"
#include "opencv2/highgui/highgui.hpp"
#include "opencv2/imgproc/imgproc.hpp"
#include <opencv2/ml/ml.hpp>
#include <iostream>
using namespace cv;
using namespace std;


//全局变量
Size sampleSize(160,160);//样本的大小
int train_samples =10;
int classes = 4;
Mat trainData;
Mat trainClasses;

//申明全局函数
Mat readImageSaveContour(Mat src);
void getData();

int main()
{
	Mat src = imread("5.png", 0);
	getData();
	//定义SVM
	/*CvSVMParams params;
	params.svm_type = CvSVM::C_SVC;
	params.kernel_type = CvSVM::LINEAR;
	params.term_crit = cvTermCriteria(CV_TERMCRIT_ITER, 100, 1e-6);*/

	CvSVMParams SVM_params;  // CvSVMParams结构用于定义基本参数  
	SVM_params.svm_type = CvSVM::C_SVC;     // SVM类型  
	SVM_params.kernel_type = CvSVM::LINEAR; // 不做映射  
	SVM_params.degree = 0;
	SVM_params.gamma = 1;
	SVM_params.coef0 = 0;
	SVM_params.C = 1;
	SVM_params.nu = 0;
	SVM_params.p = 0;
	SVM_params.term_crit = cvTermCriteria(CV_TERMCRIT_ITER, 1000, 0.01);

	CvSVM SVM;
	SVM.train(trainData, trainClasses, Mat(), Mat(), SVM_params);

	//得到图像进行预测
	Mat show;
	cvtColor(src,show,8);//8 表示灰度图到彩色图
	Mat imageWhite;
	threshold(src, imageWhite, 100, 255, 8);
	imageWhite = 255 - imageWhite;
	vector<vector<Point> > contours;
	vector<Vec4i> hierarchy;
	findContours(imageWhite, contours, hierarchy, RETR_EXTERNAL, CHAIN_APPROX_NONE);
	for (int index = contours.size() - 1; index >= 0; index--)
	{
		Rect rectangleTem = boundingRect(contours[index]);
		Mat image;
		image = Mat::zeros(src.size(), CV_8UC1);
		drawContours(image, contours, index, Scalar(255), 2, 8, hierarchy);
		Mat tem = image(rectangleTem);
		Mat imageNewSize;
		resize(tem, imageNewSize, sampleSize, CV_INTER_LINEAR);
		image.release();
		image = imageNewSize.reshape(1, 1);
		image.convertTo(image, CV_32FC1);
		int response = (int)SVM.predict(image);
		if (response == 0)
		{
			cout << "    circle" << endl;
			string str = "circle";
			putText(show, str, Point(rectangleTem.x   rectangleTem.width / 2, rectangleTem.y   rectangleTem.height / 2),
				FONT_HERSHEY_SIMPLEX, 0.5, Scalar(0, 0, 255), 1, 8);
		}
		else if (response == 1)
		{
			cout << "    rectangle" << endl;
			string str = "rectangle";
			putText(show, str, Point(rectangleTem.x   rectangleTem.width / 2, rectangleTem.y   rectangleTem.height / 2),
				FONT_HERSHEY_SIMPLEX, 0.5, Scalar(0, 0, 255), 1, 8);
		}	
		else if (response == 2)
		{
			cout << "    triangle" << endl;
			string str = "triangle";
			putText(show, str, Point(rectangleTem.x   rectangleTem.width / 2, rectangleTem.y   rectangleTem.height / 2),
				FONT_HERSHEY_SIMPLEX, 0.5, Scalar(0, 0, 255), 1, 8);
		}
		else if (response == 3)
		{
			cout << "    cross" << endl;
			string str = "cross";
			putText(show, str, Point(rectangleTem.x   rectangleTem.width / 2, rectangleTem.y   rectangleTem.height / 2),
				FONT_HERSHEY_SIMPLEX, 0.5, Scalar(0, 0, 255), 1, 8);
		}
	}
	imshow("result",show);
	imwrite("result.png", show);

	waitKey(0);
	return 0;
}


void getData()
{
	trainData.create(train_samples*classes, sampleSize.width*sampleSize.height, CV_32FC1);
	trainClasses.create(train_samples*classes, 1, CV_32FC1);
	Mat src_image;
	char file[255];
	int i, j;
	for (i = 0; i<classes; i  )
	{
		for (j = 0; j< train_samples; j  )
		{
			sprintf(file, "samples/s%d/%d.png", i, j);
			src_image = imread(file, 0);
			if (src_image.empty())
			{
				printf("Error: Cant load image %s\n", file);
				//exit(-1);
			}
			Mat image = readImageSaveContour(src_image);
			Mat imageNewSize;
			resize(image, imageNewSize, sampleSize, CV_INTER_LINEAR);
			image.release();
			image = imageNewSize.reshape(1, 1);
			image.convertTo(trainData(Range(i*train_samples   j, i*train_samples   j   1), Range(0, trainData.cols)), CV_32FC1);
			trainClasses.at<float>(i*train_samples   j, 0) = i;
		}
	}
}

Mat readImageSaveContour(Mat src)
{
	Mat imageWhite;
	threshold(src, imageWhite, 100, 255, 8);
	imageWhite = 255 - imageWhite;
	vector<vector<Point> > contours;
	vector<Vec4i> hierarchy;
	findContours(imageWhite, contours, hierarchy, RETR_EXTERNAL, CHAIN_APPROX_NONE);
	//最大轮廓
	double maxarea = 0;
	int maxAreaIdx = 0;
	for (int index = contours.size() - 1; index >= 0; index--)
	{
		double tmparea = fabs(contourArea(contours[index]));
		if (tmparea>maxarea)
		{
			maxarea = tmparea;
			maxAreaIdx = index;
		}
	}
	Rect rectangleTem = boundingRect(contours[maxAreaIdx]);
	Mat image;
	image = Mat::zeros(src.size(), CV_8UC1);
	drawContours(image, contours, 0, Scalar(255), 2, 8, hierarchy);
	//Rect newRectangleTem(rectangleTem.x - 1, rectangleTem.y - 1, rectangleTem.width   2, rectangleTem.height 2);
	Mat tem = image(rectangleTem);
	return tem;
}