科研之友
微信
新浪微博
Facebook
分享链接
摘要
Macular region is the most sensitive area surrounded by vessels. Macular vessel segmentation is helpful for measure the vessel density, whose variation is significant for diagnosing macular diseases. In this paper, a macular vessel segmentation method based on direction tracking and pixel convolution is proposed. Firstly, composite Gamma correction is used to enhance the contrast in the green channel of RGB model. Shade correction and anisotropic coupled diffusion filtering are applied to remove the avascular area and eliminate noise, respectively. Then, phase congruency is utilized to roughly extract vessels after macular region extraction using existing methods. Lastly, direction tracking based on local gradient direction analysis is proposed to connect discontinuous vessels. Based on the Gaussian-like characteristic of vessels, pixel convolution is designed to calculate the width of vessels. The segmentation results are used for vessel density measurement. The method was tested on the DRIVE and STARE databases. The experimental results demonstrate that the proposed method can segment macular vessels accurately.
