Segmenting a low contrast image

Hi, I am trying to segment a low contrast image. I have so far tried, calculating the image gradients to look at the change in pixel intensities in x y direction. But i am not able to decipher ,how to carry on from here. I have attached the image for a better look.

Please provide some solution or suggestions on how to proceed on it.

http://postimage.org/image/9r09qj7zt/9670c890/

Thanks

[kittu]

Can any 1 help me with this? I want to segment the lines and the coffee bean type structures..I need the output,something like:

http://postimage.org/image/4s0egatvj/4a8cf8a0/

Both the images might not coincide, as i have cropped and showed a smaller part of the image.

Answer 1

[Image Analyst]

I have no idea what you want to find in the image. Do you want to measure the squiggles, or the things that look like coffee beans? And what characteristic of them do you want to measure? Area fraction, major axis length, tortuosity, total perimeter length, what???

[kittu]

Sorry for the confusion. I want to segment the lines and the coffee bean type structures..I need the output,something like:

http://postimage.org/image/4s0egatvj/4a8cf8a0/

Both the images might not coincide, as i have cropped and showed a smaller part of the image.

[Image Analyst]

Did you try edge(), an edge detection function in the Image Processing Toolbox?

Answer 2

[UJJWAL]

@Kittu

Please ask the question in a clear language that makes sense to all. I understand that you want to do some kind of segmentation but it is equally important to specify that what is it really, which you want to segment out and what properties are of interest to you... Without such information, i am afraid it is not possible to help you out.