opencv3机器学习之EM算法示例详解

bool trainE(InputArray samples, InputArray means0,
                        InputArray covs0=noArray(),
                        InputArray weights0=noArray(),
                        OutputArray logLikelihoods=noArray(),
                        OutputArray labels=noArray(),
                        OutputArray probs=noArray())

bool trainM(InputArray samples, InputArray probs0,
                        OutputArray logLikelihoods=noArray(),
                        OutputArray labels=noArray(),
                        OutputArray probs=noArray())

Vec2d predict2(InputArray sample, OutputArray probs) const

#include "stdafx.h"
#include "opencv2/opencv.hpp"
#include <iostream>
using namespace std;
using namespace cv;
using namespace cv::ml;
//使用EM算法实现样本的聚类及预测
int main()
{
    const int N = 4;    //分成4类
    const int N1 = (int)sqrt((double)N);
    //定义四种颜色，每一类用一种颜色表示
    const Scalar colors[] =
    {
        Scalar(0, 0, 255), Scalar(0, 255, 0),
        Scalar(0, 255, 255), Scalar(255, 255, 0)
    };
    int i, j;
    int nsamples = 100;   //100个样本点
    Mat samples(nsamples, 2, CV_32FC1);  //样本矩阵,100行2列，即100个坐标点    
    Mat img = Mat::zeros(Size(500, 500), CV_8UC3);  //待测数据，每一个坐标点为一个待测数据
    samples = samples.reshape(2, 0);
    //循环生成四个类别样本数据，共样本100个，每类样本25个
    for (i = 0; i < N; i++)
    {
        Mat samples_part = samples.rowRange(i*nsamples / N, (i + 1)*nsamples / N);
        //设置均值
        Scalar mean(((i%N1) + 1)*img.rows / (N1 + 1),
            ((i / N1) + 1)*img.rows / (N1 + 1));
        //设置标准差
        Scalar sigma(30, 30);
        randn(samples_part, mean, sigma);  //根据均值和标准差，随机生成25个正态分布坐标点作为样本
    }
    samples = samples.reshape(1, 0);
    // 训练分类器
    Mat labels;  //标注，不需要事先知道
    Ptr<EM> em_model = EM::create();
    em_model->setClustersNumber(N);
    em_model->setCovarianceMatrixType(EM::COV_MAT_SPHERICAL);
    em_model->setTermCriteria(TermCriteria(TermCriteria::COUNT + TermCriteria::EPS, 300, 0.1));
    em_model->trainEM(samples, noArray(), labels, noArray());
    //对每个坐标点进行分类，并根据类别用不同的颜色画出
    Mat sample(1, 2, CV_32FC1);
    for (i = 0; i < img.rows; i++)
    {
        for (j = 0; j < img.cols; j++)
        {
            sample.at<float>(0) = (float)j;
            sample.at<float>(1) = (float)i;
            //predict2返回的是double值，用cvRound进行四舍五入得到整型
            //此处返回的是两个值Vec2d，取第二个值作为样本标注
            int response = cvRound(em_model->predict2(sample, noArray())[1]);
            Scalar c = colors[response];  //为不同类别设定颜色
            circle(img, Point(j, i), 1, c*0.75, FILLED);
        }
    }
    //画出样本点
    for (i = 0; i < nsamples; i++)
    {
        Point pt(cvRound(samples.at<float>(i, 0)), cvRound(samples.at<float>(i, 1)));
        circle(img, pt, 2, colors[labels.at<int>(i)], FILLED);
    }
    imshow("EM聚类结果", img);
    waitKey(0);
    return 0;
}

#include "stdafx.h"
#include "opencv2/opencv.hpp"
#include <iostream>
using namespace std;
using namespace cv;
using namespace cv::ml;
int main()
{
    const int MAX_CLUSTERS = 5;
    Vec3b colorTab[] =
    {
        Vec3b(0, 0, 255),
        Vec3b(0, 255, 0),
        Vec3b(255, 100, 100),
        Vec3b(255, 0, 255),
        Vec3b(0, 255, 255)
    };
    Mat data, labels;
    Mat pic = imread("d:/woman.png");
    for (int i = 0; i < pic.rows; i++)
    for (int j = 0; j < pic.cols; j++)
    {
        Vec3b point = pic.at<Vec3b>(i, j);
        Mat tmp = (Mat_<float>(1, 3) << point[0], point[1], point[2]);
        data.push_back(tmp);
    }
    int N =3;  //聚成3类
    Ptr<EM> em_model = EM::create();
    em_model->setClustersNumber(N);
    em_model->setCovarianceMatrixType(EM::COV_MAT_SPHERICAL);
    em_model->setTermCriteria(TermCriteria(TermCriteria::COUNT + TermCriteria::EPS, 300, 0.1));
    em_model->trainEM(data, noArray(), labels, noArray());
    int n = 0;
    //显示聚类结果，不同的类别用不同的颜色显示
    for (int i = 0; i < pic.rows; i++)
    for (int j = 0; j < pic.cols; j++)
    {
        int clusterIdx = labels.at<int>(n);
        pic.at<Vec3b>(i, j) = colorTab[clusterIdx];
        n++;
    }
    imshow("pic", pic);
    waitKey(0);
    return 0;
}