基于OpenCV的人脸签到系统

效果图

在这里插入图片描述

目录文件

在这里插入图片描述

camerathread.h 功能实现全写在.h里了

class CameraThread : public QThread
{Q_OBJECT
public:CameraThread(){//打开序号为0的摄像头m_cap.open(0);if (!m_cap.isOpened()) {qDebug() << "Error: Cannot open camera";}//判断是否有文件,人脸识别模型，检测和识别用的if(!m_cascada.load("D:/research/CV/Opencv/haarcascade_frontalface_alt2.xml")){QMessageBox::information(NULL,"失败", "人脸识别模型装载失败");}//实例化定时器，子线程中数据传输到中线程并显示出来m_timer = new QTimer(this);//绑定时间信号及获取图像帧的图像connect(m_timer,SIGNAL(timeout()),this,SLOT(readFarme()));//大概每秒24帧//开始定时器m_timer->start(42);//打卡模块// 定义 FisherFaceRecognizer 模型，训练用的模型m_model = LBPHFaceRecognizer::create();// 加载训练好的模型，自己训练的模型m_model->read("MyFaceLBPHModel.xml");if (m_model.empty()){qDebug() << "Error: Failed to load model!";}else{qDebug() << "Model loaded successfully!";}//录入时候的定时器，录入大概几秒，获取二十张灰色图像并保存，等待训练//实例化定时器m_Very_timer = new QTimer(this);//数据库的初始化部分//链接数据库m_db = QSqlDatabase::addDatabase("QMYSQL");m_db.setHostName("localhost");  // 主机名m_db.setDatabaseName("face");  // 数据库名m_db.setUserName("root");  // 用户名m_db.setPassword("31415926");  // 密码}~CameraThread(){//释放摄像头m_cap.release();}void run() override{}//图像数据类型转换QImage MatImageToQt(const cv::Mat &src){if(src.type() == CV_8UC1){QImage qImage(src.cols,src.rows,QImage::Format_Indexed8);qImage.setColorCount(256);for(int i = 0; i < 256; i ++){qImage.setColor(i,qRgb(i,i,i));}uchar *pSrc = src.data;for(int row = 0; row < src.rows; row ++){uchar *pDest = qImage.scanLine(row);memcmp(pDest,pSrc,src.cols);pSrc += src.step;}return qImage;}else if(src.type() == CV_8UC3){const uchar *pSrc = (const uchar*)src.data;QImage qImage(pSrc,src.cols,src.rows,src.step,QImage::Format_RGB888);return qImage.rgbSwapped();}else if(src.type() == CV_8UC4){const uchar *pSrc = (const uchar*)src.data;QImage qImage(pSrc, src.cols, src.rows, src.step, QImage::Format_ARGB32);return qImage.copy();}else{return QImage();}}//人脸检测void Check(Mat &image, Mat &gray){//直方图均匀化（改善图像的对比度和亮度）Mat equalizedImg;equalizeHist(gray,equalizedImg);int flags = CASCADE_SCALE_IMAGE; //检测多个人Size minFeatureSize(30,30);float searchScaleFactor = 1.1f; //默认1.1倍int minNeighbors = 4;m_cascada.detectMultiScale(equalizedImg,m_faces,searchScaleFactor,minNeighbors,flags,minFeatureSize);m_current_people = m_faces.size();//检测到的个数//qDebug() << "检测到人脸的个数：" << m_faces.size() << endl;QString str;str.setNum(m_faces.size());//qDebug() << m_current_people << endl;//        //画矩形框Mat face;
//        for(int i = 0; i < m_faces.size(); i++)
//        {
//            if(m_faces[i].height > 0 && m_faces[i].width >0 )
//            {
//                face = gray(m_faces[i]);
//                m_text_lb = Point(m_faces[i].x,m_faces[i].y);
//                rectangle(image, m_faces[i], Scalar(50, 50, 150), 2, 8, 0); //线太细了会导致在QLabel上面丢失线框//            }
//            int iP = Predict(image);//        }// 画矩形框和显示姓名for(int i = 0; i < m_faces.size(); i++){string name; // 从数据库中获取姓名if(m_faces[i].height > 0 && m_faces[i].width >0 ){face = gray(m_faces[i]);m_text_lb = Point(m_faces[i].x,m_faces[i].y);rectangle(image, m_faces[i], Scalar(50, 50, 150), 2, 8, 0); // 画矩形框//灰度图Mat tImagGray;cvtColor(image, tImagGray, COLOR_BGR2GRAY);// 识别人脸int id = Predict(tImagGray);if (!m_db.open()){qDebug() << "Failed to connect to database:" ;}// 执行查询语句QSqlQuery query;QString queryString = QString("SELECT name FROM staff_info WHERE num = %1").arg(id);if (!query.exec(queryString)){qDebug() << "Failed to execute query:";m_db.close();}// 处理查询结果if (query.next()){name = query.value(0).toString().toUtf8().constData();m_db.close();}else{m_db.close();name = "Unknown";}}qDebug() << name.data();// 在图像上显示姓名Point textPosition(m_faces[i].x, m_faces[i].y - 10); // 文本位置在矩形框上方一点putText(image, name.data(), textPosition, FONT_HERSHEY_SIMPLEX, 0.5, Scalar(255, 255, 255), 1, LINE_AA);}m_faces.clear();}// 裁剪出的脸部区域的图像cv::Mat cropFace(const cv::Mat& faceImg, const cv::Rect& faceRect) {// 不再进行颜色空间转换，直接使用输入的 faceImgcv::Mat frame_gray = faceImg;// 确保矩形区域在图像范围内if (faceRect.x >= 0 && faceRect.y >= 0 &&faceRect.x + faceRect.width <= frame_gray.cols &&faceRect.y + faceRect.height <= frame_gray.rows){// 使用区域选择功能提取矩形区域cv::Mat faceROI = frame_gray(faceRect).clone(); // 使用 clone() 来复制图像区域return faceROI;}else{// 处理矩形区域超出图像范围的情况// 这里可以选择合适的处理方式，比如调整矩形区域的大小或者放弃处理该脸部区域return cv::Mat();}}// 在文本文件末尾添加数据void appendToTextFile(const QString& filename, const QString& data){QFile file(filename);if (file.open(QIODevice::Append | QIODevice::Text)){QTextStream out(&file);out << data << "\n";file.close();}else{qDebug() << "Failed to open file for appending.";}}//预测功能，检测输入的图像中的人脸是否在训练集中int Predict(Mat src_image){Mat face_test;int ispredict = 0;//截取的ROI人脸尺寸调整if (src_image.rows >= 120){//改变图像大小，使用双线性差值resize(src_image, face_test, Size(92, 112));}//判断是否正确检测ROIif (!face_test.empty()){//测试图像应该是灰度图ispredict = m_model->predict(face_test);}//cout << ispredict << endl;return ispredict;}signals://向主线程传输图像void Set_image(QImage *image);//打卡之后设置ui部分和数据库部分void Send_Card_Data(QSqlDatabase *m_db,QString num);public slots://摄像头读取函数void readFarme(){//读取一帧图像m_cap.read(m_src_image) ;//处理一下数据，人脸检测//生成灰度图Mat dst_gray;cvtColor(m_src_image, dst_gray, COLOR_BGR2GRAY);Check(m_src_image, dst_gray);//转换图像数据类型QImage imag = MatImageToQt(m_src_image);//发送图像Set_image(&imag);}//执行二十次检测人脸并保存下来的功能void Verity(){//判断目前摄像头中的脸有几个if (!m_facenum.isEmpty()){//如果目前只有一个人if(m_current_people == 1){qDebug() << m_executionCount;//裁剪出人脸区域Mat faceROI = cropFace(m_src_image, m_faces[0]);if (!faceROI.empty()){// 调整裁剪后的脸部图像大小cv::Mat resizedFace;cv::resize(faceROI, resizedFace, cv::Size(92, 112));// 将调整大小后的脸部图像存储在 m_src_image 中m_src_image = resizedFace.clone();} else{qDebug() << "裁剪后的脸部图像为空！ ";}//存储的地址QString dir_str = "D:\\research\\CV\\Opencv\\facedata\\" + m_facenum + "\\" + QString::number(m_executionCount) + ".jpg";//用来判断这个地址的文件夹是否存在QString is_dir = "D:\\research\\CV\\Opencv\\facedata\\" + m_facenum;qDebug() << dir_str;QDir dir(is_dir);if (!dir.exists()){if (!dir.mkpath(".")){qDebug() << "default ";}}//扣的脸部的图像如果不为空if (!faceROI.empty()){string filename = dir_str.toStdString();// 将彩色图像转换为灰度图像cv::Mat frame_gray;cv::cvtColor(faceROI, frame_gray, cv::COLOR_BGR2GRAY);//存储图像imwrite(filename, frame_gray);//在存储训练集需要的图片的地址的txt里也更新地址QString csvsave = "D:/research/CV/Opencv/facedata/" + m_facenum + "/" + QString::number(m_executionCount) + ".jpg" + ";" + m_facenum ;//存放要训练的模型的图片的地址的txtQString csvfilename = "D:\\research\\CV\\Opencv\\at.txt";//添加数据appendToTextFile(csvfilename, csvsave);m_executionCount++; // 每次执行计数器加一} else{qDebug() << "脸部图像为空，无法写入文件！ ";}//存储20张之后if (m_executionCount >= 20){m_executionCount = 0;m_Very_timer->stop();}}}else{QMessageBox::about(NULL, "提示", "请输入工号！");m_executionCount = 0;m_Very_timer->stop();//delete m_Very_timer;}}//录入脸void Set_Verity_face(QString facenum,QString facename){m_facenum = facenum;m_facename = facename;qDebug()<< QSqlDatabase::drivers();//判断数据库是否开着if (!m_db.open()){qDebug() << "Failed to connect to database:" ;} else{qDebug() << "Connected to database!";}//查找全部，看数量以便添加idQSqlQuery countQuery("SELECT COUNT(*) FROM staff_info");if (countQuery.exec() && countQuery.next()){int rowCount = countQuery.value(0).toInt(); // 获取数据条数int newId = rowCount + 1; // 新的 ID 就是数据条数加一QSqlQuery query;query.prepare("INSERT INTO staff_info (num, id, name) VALUES (:facenum, :faceid, :facename)");query.bindValue(":facenum", m_facenum);query.bindValue(":faceid", newId); // 使用新的 IDquery.bindValue(":facename", m_facename);if (query.exec()){qDebug() << "Data inserted into database successfully!";}else{qDebug() << "Failed to insert data into database:" ;}}else{qDebug() << "Failed to retrieve row count from database:" ;}//关闭数据库m_db.close();qDebug() << "开始录入\n";//绑定时间信号及获取图像帧的图像connect(m_Very_timer, SIGNAL(timeout()), this, SLOT(Verity()));//开始定时器m_Very_timer->start(200);}//设置打卡部分void Set_Card(){Mat gray;cvtColor(m_src_image, gray, CV_BGR2GRAY); // 将输入图像转换为灰度图像vector<Rect> faces; // 存放检测到的人脸矩形的向量容器// 使用级联分类器检测人脸m_cascada.detectMultiScale(gray, faces, 1.1, 4, 0, Size(30, 30), Size(500, 500));// 遍历检测到的每张人脸for (size_t i = 0; i < faces.size(); i++){Rect faceRect = faces[i]; // 获取当前人脸的矩形框// 从灰度图像中提取当前人脸区域Mat faceROI = gray(faceRect);// 检查人脸区域是否为空if (!faceROI.empty()){// 调用 Predict 函数对人脸进行预测int temp = Predict(faceROI);qDebug() << "预测结果:" << temp;// 发送打卡数据，然后让主线程在数据库中查找Send_Card_Data(&m_db,QString::number(temp));}}faces.clear();}private://声明opencv的视频类cv::VideoCapture m_cap;//更新显示的定时器QTimer *m_timer;//录入的定时器QTimer *m_Very_timer;//声明Mat类图像变量，存储当前摄像头前的图像cv::Mat m_src_image;//检测的分类器CascadeClassifier m_cascada;//矩形框的点Point m_text_lb;//人脸个数vector<Rect> m_faces;// 声明一个录入计数器变量int m_executionCount = 0;//工号和姓名QString m_facenum;QString m_facename;//识别的模型Ptr<LBPHFaceRecognizer> m_model;//当前镜头前识别出的人数int m_current_people = 0;//数据库QSqlDatabase m_db;
};#endif // CAMERATHREAD_H

faceverify.cpp


#include "faceverify.h"
#include "ui_faceverify.h"#include <QDebug>
#include <QCamera>              //管理摄像头的大类
#include <QCameraInfo>          //管理摄像头的设备表
#include <QCameraViewfinder>    //管理摄像头显示区域
#include <QCameraImageCapture>  //管理图片
#include <QDateTime>            //管理时间
#include <QString>              //管理字符串
#include "verityface.h"#pragma execution_character_set("utf-8")FaceVerify::FaceVerify(QWidget *parent): QWidget(parent), ui(new Ui::FaceVerify)
{ui->setupUi(this);}FaceVerify::~FaceVerify()
{delete m_cthread;delete ui;
}void FaceVerify::Get_image(QImage *image)
{//设置图片大小和label的长宽一致QImage timage = image->scaled(ui->Camera->width(), ui->Camera->height(),Qt::IgnoreAspectRatio, Qt::SmoothTransformation);ui->Camera->setPixmap(QPixmap::fromImage(timage));
}//打开摄像头
void FaceVerify::on_pushButton_Open_Camera_clicked()
{if(ui->pushButton_Open_Camera->text() ==  "打开摄像头"){m_cthread = new CameraThread();connect(m_cthread, &CameraThread::Set_image, this, &FaceVerify::Get_image);ui->pushButton_Open_Camera->setText("关闭摄像头");}else{disconnect(m_cthread, &CameraThread::Set_image, this, &FaceVerify::Get_image);delete m_cthread;m_cthread = nullptr;ui->Camera->clear();ui->pushButton_Open_Camera->setText("打开摄像头");}
}//打卡
void FaceVerify::on_pushButton_Card_clicked()
{// 先断开之前的连接disconnect(this, &FaceVerify::Send_Card, m_cthread, &CameraThread::Set_Card);disconnect(m_cthread, &CameraThread::Send_Card_Data, this, &FaceVerify::Show_Data);// 连接信号和槽connect(this, &FaceVerify::Send_Card, m_cthread, &CameraThread::Set_Card);connect(m_cthread, &CameraThread::Send_Card_Data, this, &FaceVerify::Show_Data);// 发送信号emit Send_Card();}
//打卡数据显示
void FaceVerify::Show_Data(QSqlDatabase *m_db,QString num)
{if (!m_db->open()){qDebug() << "Failed to connect to database:";}else{qDebug() << "Connected to database!";// 执行检查并插入数据QSqlQuery query;query.prepare("SELECT * FROM record_info WHERE num = :facenum AND DATE(mtime) = CURDATE()");query.bindValue(":facenum", num);if (query.exec()){if (query.next()){// 如果有记录，表示今天已经签到QMessageBox::information(NULL, tr("提示 "), tr("今天已签到 "));// 读取数据库中的数据并显示在textEdit_data中QString data;QSqlRecord record = query.record();int numField = record.indexOf("num"); // 获取字段索引int timeField = record.indexOf("mtime");// 遍历查询结果do{QString num = query.value(numField).toString(); // 获取编号QString time = query.value(timeField).toDateTime().toString("yyyy-MM-dd hh:mm:ss"); // 获取时间data += "编号: " + num + ", 时间: " + time + "\n";} while (query.next());// 将数据显示在textEdit_data中ui->textEdit_data->setText(data);}else{QDateTime currentDateTime = QDateTime::currentDateTime();QString currentTimeString = currentDateTime.toString("yyyy-MM-dd hh:mm:ss");qDebug() << "Current time:" << currentTimeString;// 如果没有记录，表示今天还未签到，执行插入操作query.prepare("INSERT INTO record_info (num, id, mtime) VALUES (:facenum, :faceid, :facemtime)");query.bindValue(":facenum", num);query.bindValue(":faceid", num); // 使用相同的 num 作为 idquery.bindValue(":facemtime", currentTimeString);if (query.exec()){QMessageBox::information(NULL, tr("提示 "), tr("签到成功 "));}else{QMessageBox::critical(NULL, tr("错误 "), tr("签到失败： ") ); // 显示错误信息}}}else{QMessageBox::critical(NULL, tr("错误"), tr("查询数据失败： "));}// 关闭数据库连接m_db->close();}}//录入人脸
void FaceVerify::on_pushButton_face_clicked()
{//打开录入界面m_verifity = new VerityFace();connect(m_verifity, &VerityFace::Send_Face, this, &FaceVerify::GetFaceNum);m_verifity->show();
}//训练模型
void FaceVerify::on_pushButton_Train_clicked()
{m_train = new train();connect(m_train, &train::finished, this, &FaceVerify::Train_Finish);m_train->run();}
//模型训练完
void FaceVerify::Train_Finish()
{delete m_train;QMessageBox::information(NULL,"训练", "训练完成");
}