I am comparing an image with some existing images to find the match of that image in the existing images in matlab. How I can do that by using binary images?

Any body can help me in this regard?

Answer 1

[Image Analyst]

It depends on how they might be different and how close you want the match to be. For an exact match, try subtracting them. For scaled and rotated versions, or for similar images ("return all pictures of babies"), look up CBIR (content based image retrieval) in a search engine. CBIR methods are varied and include methods like SIFT and SURF and gamut similarity tests.

[Sadiq Shah]

sir I am working on gesture recognition. So I want to match the image with the exiting images and tries find out the gesture?

[Image Analyst]

I'm not familiar with gesture recognition algorithms but you can find them all here: http://www.visionbib.com/bibliography/contentspeople.html#Face%20Recognition,%20Detection,%20Tracking,%20Gesture%20Recognition,%20Fingerprints,%20Biometrics

[Sadiq Shah]

Sir, Any idea about Binary shape matching?

[Image Analyst]

What's to match? I thought they were all really parallelograms. You could try just computing the area with regionprops(). You could also compute the circularities (perimeter^2/(4*pi*area)). Also, see this: http://en.wikipedia.org/wiki/Morphometrics

[Sadiq Shah]

Binary shape matching of an image with the existing images; I am trying this by using the function im2bw, but confused how to use?

[Image Analyst]

No, you don't want that to do shape matching. That's just to get a binary image, which you already have. Can you show one of your binary images, and some of the ones that you're checking it against for a possible match? And what would constitute a match? Same areas? Same perimeter? Same shape but different angle? Same shape but different size?