SVM多分类 机器学习示例源码(MachineLearning)

【实例截图】

【核心代码】

//SVM多分类训练测试 
#include <opencv2/core/core.hpp>
#include <opencv2/imgproc/imgproc.hpp>
#include <opencv2/highgui/highgui.hpp>
#include <opencv2/ml/ml.hpp> 
#include <opencv2/opencv.hpp>  
#include <iostream>  
#include <fstream>  

using namespace cv;
using namespace std;
Size imageSize = Size(64, 64);

void coumputeHog(const Mat& src, vector<float> &descriptors)
{
	HOGDescriptor myHog = HOGDescriptor(imageSize, Size(16, 16), cvSize(8, 8), cvSize(8, 8), 9);
	myHog.compute(src.clone(), descriptors, Size(1, 1), Size(0, 0));

}

int main(int argc, char** argv)
{
	ifstream inLabels, inImages, inTestimage;
	inLabels.open("myImageLabels2.txt");		//训练图像标记
	inImages.open("myImageList2.txt");		//训练图像
	inTestimage.open("myImagetest3.txt");	//测试图像


	CvSVM *mySVM = new CvSVM();	//定义SVM分类器
	CvSVMParams params = CvSVMParams();	//定义SVM的参数集

	params.svm_type = CvSVM::C_SVC;		//SVM的类型：C_SVC表示SVM分类器，C_SVR表示SVM回归
	params.kernel_type = CvSVM::LINEAR;	//核函数类型: 线性核LINEAR:
	params.term_crit = cvTermCriteria(CV_TERMCRIT_ITER, 10000, 1e-10);
	//迭代的终止条件：CV_TERMCRIT_ITER---在完成最大的迭代次数之后,停止算法；10000 最大的迭代次数；1e-10要求的精确度

	vector<float> vecDescriptors;	//统计每个cell的梯度直方图（不同梯度的个数），即可形成每个cell的descriptor

#if(1) //条件编译。用常量来表示是否需要训练  
	//======================将图像和标签数据保存到容器结构中=================================
	string train_images;
	Mat original_image;
	Mat same_size;
	int train_labels;

	vector<Mat> vecImages;	//用来存放训练图像集
	vector<int> vecLabels;	//用来存放对应的标签

	while (inImages.good() && inLabels.good())
	{
		inImages >> train_images;
		inLabels >> train_labels;
		cout << "训练图像数据路径：" << train_images << endl;
		cout << "训练图像数据标签：" << train_labels << endl;

		original_image = imread(train_images);
		cv::cvtColor(original_image, original_image, CV_BGR2GRAY);
		cv::resize(original_image, same_size, imageSize);
		//imshow("original", same_size);

		vecImages.push_back(same_size);
		vecLabels.push_back(train_labels);
	}

	inImages.close();
	inLabels.close();
	//================================================================================
	Mat dataDescriptors;
	Mat dataResponse = (Mat)vecLabels;
	for (size_t i = 0; i < vecImages.size(); i  )
	{
		Mat src = vecImages[i];
		Mat tempRow;
		coumputeHog(src, vecDescriptors);	//HOG特征是方向梯度直方图特征，将得到的结果保存到vecDescriptors容器中
		if (i == 0)
		{
			dataDescriptors = Mat::zeros(vecImages.size(), vecDescriptors.size(), CV_32FC1);
			//dataDescriptors的行数等于训练图像数量，列数等于HOG特征数量的Mat数据
		}
		tempRow = ((Mat)vecDescriptors).t();	//t()表示转置运算，将列向量变为行向量
		tempRow.row(0).copyTo(dataDescriptors.row(i));	//将得到的特征复制到dataDescriptors矩阵中，
	}

	mySVM->train(dataDescriptors, dataResponse, Mat(), Mat(), params);	//训练数据
	//string svmName = to_string(88)   "_mysvm.xml";
	string svmName = "mysvm.xml";
	mySVM->save(svmName.c_str());	//c_str()将string类型转变为char*类型
#else  

	mySVM->load("mysvm.xml");
#endif  

	//=====================对测试图像数据预测=================================  
	string testPath;
	string Window_name;
	Mat test_image;
	Mat same_test;

	vector<float> imageDescriptor;
	int number = 0;

	cout << endl << endl << "=====测试实验开始=====" << endl;
	while (inTestimage.good())
	{
		inTestimage >> testPath;
		cout << "测试图像数据路径：" << testPath << endl;

		test_image = imread(testPath);
		cv::cvtColor(test_image, test_image, CV_BGR2GRAY);
		cv::resize(test_image, same_test, imageSize);

		Window_name = "test image"   cv::format("%.4d", number);
		imshow(Window_name, same_test);

		coumputeHog(same_test, imageDescriptor);
		Mat testDescriptor = Mat::zeros(1, imageDescriptor.size(), CV_32FC1);
		for (int j = 0; j < imageDescriptor.size(); j  )
		{
			testDescriptor.at<float>(0, j) = imageDescriptor[j];
		}
		float  label = mySVM->predict(testDescriptor, false);	//false表示多分类，true表示二分类问题，函数返回分类结果
		cout << "第 " << number << " 张测试图片的类别为：" << label << endl;

		number  ;
		waitKey(0);
	}

	inTestimage.close();
	delete mySVM;

	//system("pause");
	return 0;
}