This final practical module is the culmination of our studies. Here, we will apply fundamental concepts to solve real-world problems. This section will demonstrate how to capture and process a live video stream from a camera and then perform a classic computer vision task: detecting human faces in both static images and real-time video. This is an excellent opportunity to see how the building blocks we’ve learned can be assembled into something truly interactive and powerful.

9.1 Interfacing with Cameras for Real-Time Video

The VideoCapture class from the org.opencv.videoio package is the primary tool for interacting with cameras and video files. It provides a simple interface to open a camera device and capture frames from it one by one.

The process for capturing a single frame from a camera can be broken down into these steps:

1. Load the OpenCV native library: As always, this is the first step.
2. Instantiate VideoCapture: Create an object of the VideoCapture class. Passing 0 to the constructor typically opens the default system camera.
3. Capture a frame: Use the read() method, which captures the next available frame from the camera and stores it in a provided Mat object.

The following example captures a single snapshot from the camera, displays it in a JavaFX window, and saves the frame to a file.

import java.awt.image.BufferedImage;

import java.awt.image.DataBufferByte;

import java.awt.image.WritableRaster;

import javafx.application.Application;

import javafx.embed.swing.SwingFXUtils;

import javafx.scene.Group;

import javafx.scene.Scene;

import javafx.scene.image.ImageView;

import javafx.scene.image.WritableImage;

import javafx.stage.Stage;

import org.opencv.core.Core;

import org.opencv.core.Mat;

import org.opencv.imgcodecs.Imgcodecs;

import org.opencv.videoio.VideoCapture;

public class CameraSnapshotJavaFX extends Application {

private Mat matrix = null;

@Override

public void start(Stage stage) {

WritableImage writableImage = capureSnapShot();

saveImage();

ImageView imageView = new ImageView(writableImage);

imageView.setFitHeight(400);

imageView.setFitWidth(600);

imageView.setPreserveRatio(true);

Group root = new Group(imageView);

Scene scene = new Scene(root, 600, 400);

stage.setTitle(“Camera Snapshot”);

stage.setScene(scene);

stage.show();

}

public WritableImage capureSnapShot() {

System.loadLibrary(Core.NATIVE_LIBRARY_NAME);

VideoCapture capture = new VideoCapture(0);

Mat matrix = new Mat();

WritableImage writableImage = null;

if (capture.isOpened()) {

if (capture.read(matrix)) {

BufferedImage image = new BufferedImage(matrix.width(), matrix.height(), BufferedImage.TYPE_3BYTE_BGR);

WritableRaster raster = image.getRaster();

DataBufferByte dataBuffer = (DataBufferByte) raster.getDataBuffer();

byte[] data = dataBuffer.getData();

matrix.get(0, 0, data);

this.matrix = matrix;

writableImage = SwingFXUtils.toFXImage(image, null);

}

}

capture.release();

return writableImage;

}

public void saveImage() {

String file = “path/to/your/snapshot.jpg”;

Imgcodecs.imwrite(file, matrix);

}

public static void main(String args[]) {

launch(args);

}

}

9.2 Object Detection using Cascade Classifiers: Face Detection

Object detection is the task of finding instances of specific objects in an image. OpenCV provides a robust method for this using Cascade Classifiers. This technique uses pre-trained models, typically stored in XML files, that have learned the visual features of a particular object. We will use a classifier pre-trained to detect frontal faces.

Face Detection in a Static Image

The process for detecting faces in a static image involves three key steps:

1. Load the native library.
2. Instantiate CascadeClassifier: Create an object of this class, passing the path to the classifier XML file (e.g., lbpcascade_frontalface.xml) to its constructor. These XML files are included with the OpenCV distribution.
3. Detect faces with detectMultiScale(): This method scans the input Mat to find objects. Its output is a MatOfRect, which contains Rect objects, each representing the bounding box of a detected face. The parameters allow you to control the sensitivity of the detection, such as the minimum size of faces to detect.

import org.opencv.core.Core;

import org.opencv.core.Mat;

import org.opencv.core.MatOfRect;

import org.opencv.core.Point;

import org.opencv.core.Rect;

import org.opencv.core.Scalar;

import org.opencv.imgcodecs.Imgcodecs;

import org.opencv.imgproc.Imgproc;

import org.opencv.objdetect.CascadeClassifier;

public class FaceDetectionImage {

public static void main (String[] args) {

System.loadLibrary( Core.NATIVE_LIBRARY_NAME );

String imagePath = “path/to/your/face_image.jpg”;

Mat src = Imgcodecs.imread(imagePath);

String xmlFile = “path/to/your/opencv/sources/data/lbpcascades/lbpcascade_frontalface.xml”;

CascadeClassifier classifier = new CascadeClassifier(xmlFile);

MatOfRect faceDetections = new MatOfRect();

classifier.detectMultiScale(src, faceDetections);

System.out.println(String.format(“Detected %s faces”, faceDetections.toArray().length));

for (Rect rect : faceDetections.toArray()) {

Imgproc.rectangle(

src,

new Point(rect.x, rect.y),

new Point(rect.x + rect.width, rect.y + rect.height),

new Scalar(0, 0, 255),

3

);

}

Imgcodecs.imwrite(“path/to/your/face_detection_output.jpg”, src);

System.out.println(“Image Processed”);

}

}

The logic here is straightforward: after loading the pre-trained classifier and the image, detectMultiScale finds the faces. We then iterate through the returned rectangles and use Imgproc.rectangle() to draw a red box around each detected face on the original image before saving the result.

Face Detection in Live Video

To perform face detection on a live video stream, we simply combine the techniques from the previous two sections. We create a main loop that continuously:

1. Captures a frame from the camera using VideoCapture.
2. Runs the detectMultiScale() method on the captured frame.
3. Iterates through the resulting MatOfRect and draws rectangles on the frame.
4. Displays the annotated frame in a GUI window.

import java.awt.image.BufferedImage;

import java.awt.image.DataBufferByte;

import java.awt.image.WritableRaster;

import javafx.application.Application;

import javafx.embed.swing.SwingFXUtils;

import javafx.scene.Group;

import javafx.scene.Scene;

import javafx.scene.image.ImageView;

import javafx.scene.image.WritableImage;

import javafx.stage.Stage;

import org.opencv.core.Core;

import org.opencv.core.Mat;

import org.opencv.core.MatOfRect;

import org.opencv.core.Point;

import org.opencv.core.Rect;

import org.opencv.core.Scalar;

import org.opencv.imgproc.Imgproc;

import org.opencv.objdetect.CascadeClassifier;

import org.opencv.videoio.VideoCapture;

public class FaceDetectionJavaFX extends Application {

private Mat matrix = null;

@Override

public void start(Stage stage) {

WritableImage writableImage = captureFrame();

ImageView imageView = new ImageView(writableImage);

imageView.setFitHeight(400);

imageView.setFitWidth(600);

imageView.setPreserveRatio(true);

Group root = new Group(imageView);

Scene scene = new Scene(root, 600, 400);

stage.setTitle(“Live Face Detection”);

stage.setScene(scene);

stage.show();

}

public WritableImage captureFrame() {

System.loadLibrary(Core.NATIVE_LIBRARY_NAME);

VideoCapture capture = new VideoCapture(0);

Mat matrix = new Mat();

WritableImage writableImage = null;

if (capture.read(matrix)) {

String xmlFile = “path/to/your/opencv/sources/data/lbpcascades/lbpcascade_frontalface.xml”;

CascadeClassifier classifier = new CascadeClassifier(xmlFile);

MatOfRect faceDetections = new MatOfRect();

classifier.detectMultiScale(matrix, faceDetections);

System.out.println(String.format(“Detected %s faces”, faceDetections.toArray().length));

for (Rect rect : faceDetections.toArray()) {

Imgproc.rectangle(

matrix,

new Point(rect.x, rect.y),

new Point(rect.x + rect.width, rect.y + rect.height),

new Scalar(0, 0, 255),

3

);

}

BufferedImage image = new BufferedImage(matrix.width(), matrix.height(), BufferedImage.TYPE_3BYTE_BGR);

WritableRaster raster = image.getRaster();

DataBufferByte dataBuffer = (DataBufferByte) raster.getDataBuffer();

byte[] data = dataBuffer.getData();

matrix.get(0, 0, data);

this.matrix = matrix;

writableImage = SwingFXUtils.toFXImage(image, null);

}

capture.release();

return writableImage;

}

public static void main(String args[]) {

launch(args);

}

}

The power of combining these fundamental techniques allows for the creation of sophisticated and interactive computer vision applications.