@Override
public void run() {
// Initialisation
cptRect = 0;
initialiseRectangle();
if (video.isOpened()) {
while (begin == true) {
// On récupère l'image de la CaptureVideo
video.retrieve(frameaux);
// On modifie les dimensions de la frame
Imgproc.resize(frameaux, frame, frame.size());
// On copie
frame.copyTo(currentFrame);
if (jCheckBoxMotionDetection.isSelected()) {
if (firstFrame) {
frame.copyTo(lastFrame);
firstFrame = false;
continue;
}
// Soustraction de currentFrame par rapport à la dernière
Core.subtract(currentFrame, lastFrame, processedFrame);
// Filtre en niveau de gris
Imgproc.cvtColor(processedFrame, processedFrame, Imgproc.COLOR_RGB2GRAY);
// Filtre threshold + récupération du Jslider
int threshold = jSliderThreshold.getValue();
Imgproc.threshold(
processedFrame, processedFrame, threshold, 255, Imgproc.THRESH_BINARY);
// Detecte les contours et check dans les
detection_contours(currentFrame, processedFrame);
}
// Dessine les rectangles d'authentifications
drawRectangle();
currentFrame.copyTo(processedFrame);
// Encodage de la frame en MatOfByte
Highgui.imencode(".jpg", processedFrame, matOfByte);
byte[] byteArray = matOfByte.toArray();
// Affichage de l'image
try {
in = new ByteArrayInputStream(byteArray);
bufImage = ImageIO.read(in);
image.updateImage(bufImage);
} catch (Exception ex) {
ex.printStackTrace();
}
try {
Thread.sleep(50);
} catch (Exception ex) {
ex.printStackTrace();
}
}
}
}
// Bhattacharyya coefficient for two distributions
// a, b are original data samples
// better if a, b are dimension-reduced Principal Components
public static double Bhattacharyya(Mat a, Mat b) {
// if not dimension-reduced
// int k = a.cols()/2;
// Mat temp1 = new Mat();
// Mat temp2 = new Mat();
// Mat pa = new Mat();
// Mat pb = new Mat();
// Core.PCACompute(a, temp1, temp2, k);
// Core.PCAProject(a, temp1, temp2, pa);
// Core.PCACompute(b, temp1, temp2, k);
// Core.PCAProject(b, temp1, temp2, pb);
//
Mat cova = new Mat();
Mat covb = new Mat();
Mat meana = new Mat();
Mat meanb = new Mat();
Core.calcCovarMatrix(a, cova, meana, Core.COVAR_NORMAL | Core.COVAR_ROWS);
Core.calcCovarMatrix(b, covb, meanb, Core.COVAR_NORMAL | Core.COVAR_ROWS);
// if not dimension-reduced
// Core.calcCovarMatrix(pa, cova, meana, Core.COVAR_NORMAL|Core.COVAR_ROWS);
// Core.calcCovarMatrix(pb, covb, meanb, Core.COVAR_NORMAL|Core.COVAR_ROWS);
// System.out.println("cov");
// System.out.println(cova.dump());
// System.out.println(covb.dump());
double detcova = Core.determinant(cova);
double detcovb = Core.determinant(covb);
Mat cov = new Mat();
Core.add(cova, covb, cov);
// cov = cov.mul(cov, 0.5);
// Core.scaleAdd(cov, 0.5, new Mat(), cov);
for (int i = 0; i < cov.rows(); i++)
for (int j = 0; j < cov.cols(); j++) {
cov.put(i, j, 0.5 * cov.get(i, j)[0]);
}
double detcov = Core.determinant(cov);
Mat inv = new Mat(); // inverse of cov
Core.invert(cov, inv);
Mat mabsub = new Mat();
Core.subtract(meana, meanb, mabsub);
Mat mabsub_tp = new Mat();
Core.transpose(mabsub, mabsub_tp);
Mat temp = new Mat();
Core.gemm(mabsub, inv, 1, new Mat(), 0, temp);
// System.out.println(temp.dump());
// System.out.println(mabsub_tp.dump());
Mat res = new Mat();
Core.gemm(temp, mabsub_tp, 1, new Mat(), 0, res);
// System.out.println(res.dump());
// Mat temp = mabsub_tp.mul(inv);
// System.out.println(temp.dump());
//
// Mat res = temp.mul(mabsub);
// System.out.println(res.dump());
double s2 = Math.sqrt(detcov / (Math.sqrt(detcova * detcovb)));
// s1 = 0.125 * (mabsub_tp*inv*mabsub)
double s1 = 0.125 * res.get(0, 0)[0];
double bdis = 1 / (Math.exp(s1) * s2);
return bdis;
}