The image is basically a grayscale image having pixel intensities in the range 0-226. I have done image processing by converting this gray scale image into binary image using OTSU algorithm and got the threshold value intensity as 127.0. Basically, this OTSU works for bimodal images. But I saw in many texts stating that bimodal images are something which has some pattern like Gaussian or Skewed or uniform kind of histogram distributions etc.
Vangety Nikhil Dear Sir, I don't see any gaussian or uniform or skewed distribution. I can see the pixel intensities are randomly distributed as I see many large spikes at pixel intensities around 75 and 230. If that is because of the noise smooth it and the check the histogram.
Sandhanakrishnan Ravichandran , Thank you sir for your valuable suggestion.
Sandhanakrishnan Ravichandran, now I have done Image smoothing by Gaussian blur and histogram looks as the below attachment. What are your comments on this? Is this now unimodal or bimodal?
Vangety Nikhil Yes Sir now it is more like a bimodal distribution. I see two modes here.
Sandhanakrishnan Ravichandran Sir, using OTSU algorithm for image thresholding I got for the above smoothed version of histogram , the threshold value as 107.0. But according to the algorithm, threshold value should be between the modes , right? So, my doubt is OTSU algorithm working fine here?
Hi,
I suggest you create a histogram before the binarization step. Indeed, the average of the histogram will allow you to define the best threshold to be applied during the binarization process.
Good luck with the test.
- Badges
- Science topic
- Similar topics
- Computer Science
- Images