Unexpected segmentation of image

Hello everybody,

I have following image which I want to segment:

Here is the result I have so far:

However my expected result should look as in the following image:

(ROI).

The algorithm I have in my mind right now is:

1. Use morpholgy operation to separate noise from ROI.
2. Find biggest and brightest blows.
3. Use bwconvhull() and then bwboundaries() to find outer and inner contour of the image. “The outer is the boundaries of the convex hull. To get the inner, subtract or XOR the original from the convex hull, then call bwboundaries().” © Image Analyst

But the above algorithm may not work due to the large noise.

If you have any suggestions how I can segmented the Regions of Interes (ROI), please don't hesitate to leave a comments.

Answer 1

[Image Analyst]

I have no idea what it is about that part of the image that distinguishes it from any other part of the image. How does one know that the upper white stuff is not desired and the lower white stuff is desired? And how do you decide the dividing line between those two regions?