Identifying and isolating brighter parts in an image

Hi everyone,

I need to identify and isolate the brighter parts in the attached image. My problem is that I have to do this operation on a large amount of images, similar to this one, but with different intesities of the brighter parts and of the background. Therefore I wanted to find a way to automate the process, but if I calculate the value of thresholding with graythresh and then use im2bw with it I get too many white parts. I should use a costumed value of thresholding, but that varies depending on the image and I cannot automate the process... Any idea?

Rita

[Andres Fonts]

Hi Ritird, I did some image processing in the past, and I remember to find similar problems.

Have you tried to use edge-detection filters? It may help you. For instance, first you apply an edge detection filter to the image, and then you find the threshold. You may need to play with the threshold and also with the kind of filter (roberts, sobel, kirsh...).

Sorry if this cannot help you!

Answer 1

[Image Analyst]

I don't know what identify and isolate means to you. Please clarify. Perhaps it means that you simple threshold and get a binary image. Perhaps you want the centroids. Perhaps you want to crop each blob out into a new, smaller image. I don't know so please tell us.

You might be interested in my Image Processing Tutorial, or my interactive thresholding app in my File Exchange: http://www.mathworks.com/matlabcentral/fileexchange/?term=authorid%3A31862

[Ritird]

By identifying and isolating the brighter parts I meant thresholding and getting a binary image where the brighter parts are white and everything else is black.

[Image Analyst]

Then threshold:

binaryImage = grayScaleImage > someThreshold;