How to detect certain shapes in binary image

I have a binary image containing circular shape; I want to detect the circular shape and eliminate rest of the part in an image. Can someone help me with this. The original image is attached below http://img841.imageshack.us/img841/9128/originalimage.jpg

Expected output image (Edited in MS-Paint)

http://img856.imageshack.us/img856/1872/expectedoutput.jpg

Thanks for your help in advance!

[Matt Kindig]

If you have the Image Processing Toolbox, I would go through the Demos, specifically the one called "Identifying Round Objects". Another excellent resource is ImageAnalyst's BlobsDemo on the File Exchange ( http://www.mathworks.com/matlabcentral/fileexchange/25157-image-segmentation-tutorial-blobsdemo). I would look at both to start with.

Answer 1

[Image Analyst]

If you look at the circularity, that will tell you what's circular and what's not. Look at my BlobsDemo and see how to measure area and perimeter. Then circularity is just perimeter^2 / (4*pi*area). It's 1 for a circle, and gets bigger the more tortuous the edge shape gets.

[John]

The link is broken

[Image Analyst]

The link is not broken. I just went there and it works fine for me. Make sure you went to the right person's File Exchange (i.e., mine). Here is the direct link: http://www.mathworks.com/matlabcentral/fileexchange/25157-image-segmentation-tutorial-blobsdemo