Project 3: Feature Detection

Blob Detection and User assissted ROI option.

I am assumming that you will be first doing Blob detection in your breast image (if you propose something different you need to convince me of it and it will need to be approved). This blob detection algorithm is part of the work of this project. To make things easier, I will allow you to require the user to first signify an area of interest, which we will henceforth call Region of Interest (ROI), by in your applications GUI window containing the image lettting the user with their most click and drag to create a rectangle (you already were to have implemented this in your Project 2 work). This area the must be displayed separately as the active image (in essence cropping) and then you can do the processing on this. The idea is that the user has assisted with indicating a smaller area that might contain a blob. This can significantly make your blob detection easier with simpler methods. It is not required that you do the ROI option. Be specific in your proposal if you are going to require the user to do a ROI...but, note that a user could choose to do a larger area than you might think or even an ROI that doesn't contain any blobs.

Note: you may want to have a filter saying if the blob is smaller (# pixels in blob) than some ammount it is too small to be considered a Mass.

 



Here we see the user
with their mouse specifying
the region of interest- red
rectangle - that is then shown on right. This is the image that will now be processed by the system.

 

 

 

---

One Possible Idea

Given things you have learned in this class (and some implemented in previous image processing project):

step 1) Do ROI


step 2) Do thresholding. (how find threshold point? each image may need different threshold? use histogram. This is the biggest problem in this algorithm)

 

 

 

 

 

 

step 3) Group to create labeled blobs (here showing image with different blobs in different colors)...see our previous lecture on blob detection in binary images