2

I have multiple simple circle objects in grid of an image from which I want to create mask image for the objects. A gotcha is that light intensity for each object is different. So simple thresholding would not create a mask.

As a solution, I want to threshold based on gradient. Basically, I'd like to first find the circle with edge detection and make inside of the circle white and outside black. But this is really slow. Is there any better way to do this on matlab?

Tae-Sung Shin
  • 20,215
  • 33
  • 138
  • 240
  • Hough transforms? like: http://www.mathworks.com/help/toolbox/images/ref/imfindcircles.html – Ashish Uthama Jun 11 '12 at 20:31
  • Show an image if you please. In the mean while have a look at this [DSP](http://dsp.stackexchange.com/questions/1932/what-are-the-best-algorithms-for-document-image-thresholding-in-this-example/1934#1934) and [SO](http://stackoverflow.com/questions/9937316/document-image-binarization/9940718#9940718) question for inspiration on local background estimation. – Maurits Jun 12 '12 at 07:43

1 Answers1

2

I would create a low-pass filtered version of the image, and use it as the threshold. The "strength" of the filter should be tuned carefully in order to make the result follow the distribution of light intensity, but this is not that hard.

(This approach worked for me when I had to extract the contour of blood vessels from brain-surface images, few years ago.)

kol
  • 27,881
  • 12
  • 83
  • 120