opencv人脸识别

谁践踏了优雅 2022-05-15 01:58 342阅读 0赞

## opencv人脸识别： ##

首先我们来简单的讲一下人脸识别流程：这里采用的是vs2017+opencv3.3扩展库

1、进行人脸检测

2、识别器训练与分类

3、人脸检测完和识别器进行预测

4、圈出人脸并在人脸上面显示人名

注：这里用的库比较多是opencv扩展库中的，官方的库会报错(因为没有相应的EigenFaceRecognizer类模板无法进行识别器的训练分类) 可以参考上一篇博客如何编译opencv扩展库  还要注意的是不同版本的识别器创建的写法会不同，最好百度查一下api（这个也是搞了我蛮久的）

Code：

//face_recog_from_video.cpp 定义控制台应用程序的入口点。
    
    #include "pch.h" //预编译 不同版本不同  
    #include "opencv2/opencv.hpp"
    #include "opencv2/face.hpp"
    #include <stdio.h>
    #include<iostream>
    
    using namespace std;
    using namespace cv;
    using namespace cv::face;
    
    /** Function Headers */
    void detectAndDisplay(Mat frame);
    
    /** Global variables */
    String face_cascade_name = "G:\\opencv\\opencv+\\opencvmake\\install\\etc\\haarcascades\\haarcascade_frontalface_alt.xml";
    CascadeClassifier face_cascade;
    
    String window_name = "Capture - Face Recognition";
    string face1_path = "D:\\xue.jpg";//1
    string face2_path = "D:\\zxy.jpg";//2
    string face3_path = "D:\\zly.jpg";//3
    string face4_path = "D:\\my.jpg";//我->4号
    
    Ptr<EigenFaceRecognizer> modelPCA = EigenFaceRecognizer::create();// 创建特征脸模型  EigenFaceRecognizer
    
    /** @function main */
    int main(int argc, const char** argv)
    {
    
    	vector<Mat> images;
    	images.push_back(imread(face1_path, CV_LOAD_IMAGE_GRAYSCALE));
    	images.push_back(imread(face2_path, CV_LOAD_IMAGE_GRAYSCALE));
    	images.push_back(imread(face3_path, CV_LOAD_IMAGE_GRAYSCALE));
    	images.push_back(imread(face4_path, CV_LOAD_IMAGE_GRAYSCALE));
    
    	vector<int> labels(4);
    	for (int i = 1;i <= 4;i++)
    		labels[i - 1] = i;
    
    	//训练
    	modelPCA->train(images, labels);
    
    	VideoCapture capture;
    	Mat frame;
    
    	//-- 1. Load the cascades
    	if (!face_cascade.load(face_cascade_name)) { printf("--(!)Error loading face cascade\n"); return -1; };
    
    	//-- 2. Read the video stream
    	capture.open(0); //打开摄像头
    	if (!capture.isOpened()) { printf("--(!)Error opening video capture\n"); return -1; }
    
    	while (capture.read(frame)) //读取帧
    	{
    		if (frame.empty())
    		{
    			printf(" --(!) No captured frame -- Break!");
    			break;
    		}
    
    		//-- 3. Apply the classifier to the frame
    		detectAndDisplay(frame);
    
    		if (waitKey(10) == 'k') { break; } // escape  
    	}
    	return 0;
    }
    
    /** @function detectAndDisplay */
    void detectAndDisplay(Mat frame)
    {
    	std::vector<Rect> faces;
    	Mat frame_gray;
    
    	cvtColor(frame, frame_gray, COLOR_BGR2GRAY);  //BGR 转化为灰度图
    	equalizeHist(frame_gray, frame_gray);   //直方图均衡化
    	//-- Detect faces
    	face_cascade.detectMultiScale(frame_gray, faces, 1.1, 2, 0 | CASCADE_SCALE_IMAGE, Size(60, 60));
    
    	for (size_t i = 0; i < faces.size(); i++)
    	{
    		Point center(faces[i].x + faces[i].width / 2, faces[i].y + faces[i].height / 2); // 人脸中心坐标
    		ellipse(frame, center, Size(faces[i].width / 2, faces[i].height / 2), 0, 0, 360, Scalar(255, 0, 255), 4, 8, 0); // 椭圆
    
    		Mat faceROI = frame_gray(faces[i]);
    		Mat face_test;
    		// 调整大小为92*112  
    		resize(faceROI, face_test, Size(92, 112));
    		imwrite("lisa.pgm", face_test);
    		double confidence;
    		int predictPCA;
    		modelPCA->predict(face_test, predictPCA, confidence);
    		cout << "the predict result is " << predictPCA << endl << "confidence is " << confidence << endl;
    		if (predictPCA == 4)
    		{
    			putText(frame, "Hello lzw", Point(faces[i].x, faces[i].y), FONT_HERSHEY_SIMPLEX, 1.5, Scalar(0, 0, 255), 2);
    
    		}
    
    
    	}
    	//-- Show what you got
    	imshow(window_name, frame);
    }

![70][]

哈哈哈~~~~~~~图片被做了一点处理，大致就是这样子啦

参考博客：[https://blog.csdn.net/u012679707/article/details/80410094][https_blog.csdn.net_u012679707_article_details_80410094]

这里还是有个问题，图片的识别器训练与分类需要在代码中实现，如果图片数目比较多的话 工作量就会比较大而且也比较耗时。

下面是用csv文件的形式实现识别器的训练与分类 可以很方便的处理图片数目多的训练

[https://blog.csdn.net/u012679707/article/details/80397012][https_blog.csdn.net_u012679707_article_details_80397012]

[70]: /images/20220515/3607c8fcba174374854da1f9db8d7cba.png
[https_blog.csdn.net_u012679707_article_details_80410094]: https://blog.csdn.net/u012679707/article/details/80410094
[https_blog.csdn.net_u012679707_article_details_80397012]: https://blog.csdn.net/u012679707/article/details/80397012