November Week 4

The group has been testing the different thresholding methods on different image masks , and have been trying to decide which method works best. To do so, we thought it was necessary to develop an evaluation metric for our images. First, we plan on using the DRIVE database for our groundtruth as before. So, whatever images we test, we can compare them to the DRIVE database’s groundtruth for the blood vessel detection. While there’s no true 100% accuracy, we are hoping to see improvement in the accuracy of our reproduced images. Following this week, we will look at and compare different resulting masks for local, adaptthresh, and global thresholding methods to the groundtruth provided to give ourselves an adequate evaluation metric.