How to create histogram for finding the threshold for object detection?

Hi, i'm doing my project on hand gesture recognition. And trying to find out the threshold value for segmenting the hand region. Anyone, can you please tell me how to find out the threshold value using histogram based method?

Answer 1

[Image Analyst]

I'd take the histogram of your image and then plot it and look at the shape to determine how to threshold. There are many, many ways.

% Let's compute and display the histogram.
[pixelCount, grayLevels] = imhist(grayImage);
bar(grayLevels, pixelCount);
grid on;
title('Histogram of original image', 'FontSize', fontSize);
xlim([0 grayLevels(end)]); % Scale x axis manually.

[Nafeesath Musfira]

sir, i've used hist() function instead of imhist, since i'm working on a color image. but i've no idea how to find out the threshold for the hand region. what represents the x and y axis of the histogram of my rgb(or HSI) image.

[Image Analyst]

Are you asking for us to give you image processing advice blind, without an image? What can I say?

[Nafeesath Musfira]

Oops! Sorry sir. I was confused. This is my image,

[Image Analyst]

I think you're also confused about what would be the best kind of background to use. Why are you using a colored, patterned background instead of something like black velvet, or at the very least a colored background that has no pattern and a very much different color than the hand? So your first step is to fix your image capture problem. Then finding the threshold will be much, much simpler. Otherwise you'll have to do color segmentation and the patterns will not let you get a very good boundary to your hand.

[Nafeesath Musfira]

Yup, that's true sir. I'll try with simpler background. Thank you sir.