Image segmentation of a human body binary image

Hi I have a database of binary images of a person in different poses(running,walking etc)I want to segment the image into different parts hands legs etc. and also i want to compare the changes in its position of its hands and legs etc with respect to time using that database ie

I will first segment one image and then find a relation between those segmented parts of one image with the segmented parts of the next image and this process will happen in a loop and also somehow I create a relation between the changes.

How do I do this?Has this kind of segmentation done already?Is there a general approach ?Will ANN(artificial neural networks) be used for this?

Answer 1

[Image Analyst]

See section 2.14 for lots of published algorithms. http://www.visionbib.com/bibliography/contentspeople.html#Face%20Recognition,%20Detection,%20Tracking,%20Gesture%20Recognition,%20Fingerprints,%20Biometrics

Pick one and code it up.

[Newman]

dear image analyst thank you very much for your response.But I was thinking of something more like a general approach towards segmentation of binary images.beacuse the papers provided in your link are for 3d scans not binary images.So i want to ask you is there a way to segments body parts from a simple binary images and continue to do do this process for a sample set of images ?

[Image Analyst]

There probably is but I doubt it would be some simple chunk of a few hundred lines of code. You're not going to find any code that is simple if it's also to be robust to lots of different silhouettes. If it were so simple, you would not find lots of papers and decades of research on it. However if you want to do it manually, you can use imline(). I attach a demo of how you can create a mask of lines that you can then use to write the background in, essentially separating the silhouette into different regions.