comparing two images with diffusion mapping

hello my project is to compare a contour with a data set sample of same kind of shapes but to the nearest possible shape wth nearest intensity so which techniques can i use to compare the two images except size and histogram technques.

Answer 1

[Image Analyst]

Many of the measurements that regionprops() can make can be used to compare shapes (contours). Print them out and see which are similar and which are different for the shapes you want to compare.

[PK]

@Image Analyst hello sir can you please elaborate this answer with some script or example

[Image Analyst]

Several examples are in my File Exchange, particularly BobsDemo: http://www.mathworks.com/matlabcentral/fileexchange/?term=authorid%3A31862. (If that helps, Accept this answer.)

[PK]

@Image Analyst sir are there some ways to compare one image with several images in a file and display the nearest matched image, i have tried with one to one matching independently it worked but did not know how to do with a set of images at a time suggest some technique or a sample code ....

[Image Analyst]

How about just subtracting them? Or use other metrics like SSIM? Or use SURF? Or go all out and use some sophisticated CBIR methods? I don't know the "diffusions mapping" method.