Image Processing in Java - Face Detection

Prerequisites:

• Image Processing in Java - Read and Write
• Image Processing In Java - Get and Set Pixels
• Image Processing in Java - Colored Image to Grayscale Image Conversion
• Image Processing in Java - Colored Image to Negative Image Conversion
• Image Processing in Java - Colored to Red Green Blue Image Conversion
• Image Processing in Java - Colored Image to Sepia Image Conversion
• Image Processing in Java - Creating a Random Pixel Image
• Image Processing in Java - Creating a Mirror Image

Face Detection using OpenCV

In the Introductory Set on Image Processing, BufferedImage class of Java was used for processing images the applications of BufferedImage class are limited to some operations only, i.e, we can modify the R, G, B values of the given input image and produce the modified image. For complex image processing such as face/object detection, the OpenCV library is used which we will use in this article. 

At first, we need to set up OpenCV for Java, we recommend using eclipse for the same since it is easy to use and set up. Now let us understand some of the methods required for face detection. 
 

1. CascadeClassifier(): This class is used to load the trained cascaded set of faces which we will be using to detect faces for any input image.
2. Imcodecs.imread()/Imcodecs.imwrite() : These methods are used to read and write images as Mat objects which are rendered by OpenCV.
3. Imgproc.rectangle() : Used to generate rectangle box outlining faces detected, it takes four arguments - input_image, top_left_point, bottom_right_point, color_of_border.

Implementation:

Note: The front face cascade file can be downloaded from a local git repository. 

The input image is as follows:

Example:

Output:

Face Detected

The output of the program is shown below  my pic before and after face detection

Output Explanation:

• It loads the native OpenCV library to use Java API. An instance of CascadeClassifier is created, passing it the name of the file from which the classifier is loaded.
• Next, the detectMultiScale method is used on the classifier passing it the given image and MatOfRect object.
• MatOfRect is responsible to do face detections after processing.
• The process is iterated for doing all the face detections and marking the image with rectangles and at the end, the image is saved as an "output.png" file.