Prachi Arora ,

Taking the description from [1]

"The watershed transform finds "catchment basins" or "watershed ridge lines" in an image by treating it as a surface where light pixels represent high elevations and dark pixels represent low elevations. The watershed transform can be used to segment contiguous regions of interest into distinct objects"

if the attached image is the original, then the watershed is not the algorithm you need.

If the image attached is the original, I would go with density methods to segment the image.

[1]https://www.mathworks.com/help/images/ref/watershed.html

Just playing... applied a Gauss filter, which removes small dots, then a nonlinear transfer function, which is some kind of threshold. Bigger dots get to lie on the same area, if they are close to each other.

Regards,

Joachim

Hi Prachi Arora

What you are after is not clear. If what you are displaying is some kind of microscopic cultured cells/tissue in a petri dish and you are trying to quantify those "nuclei-like" then I would recommend a simple (background subtraction + thresholding using moments, see attached example), you can do this easily in the ImageJ tool.

W.r.t "If I want to segment based on area of pixels how I can do that." This implies that you do the binarisation first (i.e., area is usually attributed to a morphological operation acting on binary images).

Useful techniques which have shown promising results in the past include the SIFT and SURF which isolate potential points of interest by tracking the change in image gradients along horizontal and vertical directions.

Thanks everyone

Dr. Abbas sir thanks you tried it, can you please tell in above picture you applied thresholding . Actually what I am trying is to count the dots and remove very small cells or dots from the images.