Automatic Thresholding

Introduction

Here we’re going to segment one image ( with a bimodal histogram ), using Otsu’s Algorithm, which tries to maximize inter class variance, by minimizing intra class variance. Finally, we obtain suitable intensity value ( to be used as threshold ), for that intensity value which is having highest inter class variance.

Pixel intensity classes defined by selecting a threshold ∈ [0, 255]

• Intensity < threshold :> class one
• Intensity >= threshold :> class two

Details available here.

Usage

Image to be segmented needs to be bi-modal, only then this algorithm performs well.

• Make sure you’ve JDK installed & added in.itzmeanjan.filterit.jar to your Java project.
• Now in Main.java, put following code snippet.

// this is Main.java
// assuming weaver.jpg is present in current working 
// directory or put image file name as you wish

import in.itzmeanjan.filterit.ImportExportImage;
import in.itzmeanjan.filterit.segmentation.AutomaticThresholding;

public class Main {

    public static void main(String[] args) {
        AutomaticThresholding automaticThresholding = new AutomaticThresholding();
        System.out.println(
                ImportExportImage.exportImage(
                        automaticThresholding.segment("./weaver.jpg"),
                        "segmentedUsingAutomaticThresholding.jpg"));
    }
}

• Compile Main.java using following bash command.

$  javac -cp ".:in.itzmeanjan.filterit.jar" Main.java

• Run it with following command

$  java -cp ".:in.itzmeanjan.filterit.jar" Main

• And here’s your result.

Result

Original Image	Segmented Image

Thanking you :)