How to detect a polygon of following Shape in an image with other objects. I greatly appreciate for some help.

Answer 1

[Image Analyst]

Define exactly what does "detect" mean to you.

In hte meantime, see my Image Segmentation Tutorial: http://www.mathworks.com/matlabcentral/fileexchange/?term=authorid%3A31862

[Rohit Tripathi]

Sir First thanks for your answer, I actually want to detect a L shape polygon from a image...Image may contain many object but i want only that part of image that contain L shape Polygon.

[Image Analyst]

I still don't know what that means. Do you want a binary image of the black outlines region filled in? Or of just the black outline itself? Or of the (x,y) coordinates of the outer and inner perimeter of the black outline? Or the area of the filled object? Or the area of only the black outline itself? Or just a yes/no decision if there is something black in the image? Or something else?

[Image Analyst]

Please attach your actual image with all the other shapes.

[Rohit Tripathi]

Sir consider the following image from which i want to detect only the polygon of L shape..ie my output would be the L shape polygon with 6- vertices

Output should be like following....that is L shaped polygon detected in the image

[Image Analyst]

What can vary? Easiest is to just find purple blobs of that particular color. But if they can vary in color, that might not work. Could all the shapes be the same size? Can the L vary in aspect ratio? If not, the solidity would be a good thing to measure, and probably the next easiest.

This looks so fake that it's probably your homework assignment rather than a real world situation. What's the use case here?

[Rohit Tripathi]

Sir color can vary....and L can vary in aspect ratio too....Sir actually i want to detect real world objects which can be map to some polygon. Is there any method so that we can detect polygon of any shape ?

[Image Analyst]

Alright, I did it for you. It works for this image. Since you didn't answer the question about what can vary from image to image, I give no gaurantees that it works for other images. See code attached below image.

[Image Analyst]

Rohit, does it not do exactly what you asked for?

Answer 2

[Rushikesh Tade]

You can use Active shape model for that.

Here is one paper which seems to be solving same type problem. Download Paper

[Rohit Tripathi]

Thanks for your answer....Sir the paper u have mentioned does not exist on the link u have provide...sir please provide the paper. Thanks in advance...

[Star Strider]

See: ‘Multi-object segmentation using shape particles’

[Rushikesh Tade]

Same link as above that Star Strider has given

[Image Analyst]

Same document, except that Star's link worked. Hover over your link to discover why it didn't work. http://www.mathworks.com/matlabcentral/answers/13205-tutorial-how-to-format-your-question-with-markup