Seventh Week Progress Update

Following last week's literature review, we discovered that in blood vessel detection, the green channel is the channel that should be extracted for detection. This is simply because the green channel presents the largest contrast between the features of the image. As seen here, the blood vessels are best clearly visible in the green channel.

In addition, the article mentioned applying a median filter, which would produce the "estimated background" of a retinal image, which is essentially an image removing the blood vessels. So, if we have the original image with the blood vessels and the "estimated background image" with the blood vessels removed, the next step is to subtract the elements of the two.

Here, we took the median filter-enhanced image minus the green channel image, which resulted in the following image with solely blood vessels.

In conclusion, we are slowly concluding the blood vessel detection component of our project. In the next few weeks, we will begin the fovea and optical disk detection.