Enhancement/Detection of Vessel Trees in Retinal-Images Abstract
Retinal-images, which normally involve blood-vessel trees, can play an important role in many crucial applications, e.g., security applications (e.g., personal identification) and diagnostic applications (e.g., diseases detection). In all cases, the starting point in such applications is the detection of these vessel trees. This procedure can be seen as an edge detection procedure.
To produce such images, a special camera is used to picture the eye-fundus. Unfortunately, this picturing procedure not only produces the vascular patterns in eye-fundus, but some undesired noise patterns as well. Due to such noise, edge detection algorithms might fail to detect parts of existing vessels or might produce parts of un-existing vessels. Hence, unsatisfactory results are yielded. To achieve better edge detection results, an image enhancement procedure, which attenuates the noise signal and boosts the image signal, is needed to be applied first.
In this proposed project, the issue of vessel enhancement/detection will be investigated. The main objective of this project is to come up with specializedvessel enhancement/detection schemes. These schemes will take in their consideration the nature of vascular patterns, as well as the nature of applications. Hence, they expected to achieve not only robust edge detection results, but facilitate the retinal-image original application as well.