Research Methods Of Retinal Vessel Enhancement

Nowadays, the vascular disease has been one of the main death reasons for human being, and the mortality rose year by year. Retinal vessel is the only microvascular that can be observed by a noninvasive test. The characteristic changes of Retinal vessels can reflect not only many different eye diseases, but also the vascular disease of the human body, such as hypertension, diabetes, arteriosclerosis and so on. It can provide the reference and help for timely detection and treatment of related disease by viewing and analyzing the Retinal images. However, because of the imaging system is affected by equipments, imaging environment and correlated noises, the quality of retinal vessel image is usually poor, so that hard to used for medical diagnosis directly. To deal with the problem, this paper proposed three methods for retinal vascular enhancement on the basis of summarizing and analyzing the enhancement methods that exist:(1)Proposed a retinal image enhancement method based on multi-scale Hessian matrix. According to the geometrical structure and different width of retinal vessels, proposed a vessel enhancement method based on multi-scale Hessian matrix. And it also offered the basis for determining the range of scaling factors in this paper.(2) Proposed a retinal image enhancement method based on morphology filter and differential evolution algorithm. According to gray difference between vessels and background in the image, proposed a morphology-based homomorphic filter method. The differential evolution algorithm is applied to find the optimal parameters.(3) Proposed a retinal vessel enhancement method based on multi Scale Top-hat transformation and histogram fitting linear transformation. Firstly, the optimal bright and dim image regions and details are extracted by the top-hat transformation. Then, the retinal image is enhanced preliminarily by adding the optimal bright image regions and details on and subtracting the optimal dim image regions and details from the original retinal vessel image. Finally, according to the normal distribution characteristics of the histogram of the preliminarily enhanced result, we further enhance the retinal image by linear transformation with histogram Gaussian curve fitting.The experimental results show that the proposed method improves the contrast and enhances the details of the retinal vessels more effectively than the existing methods.