Fundus image processing and feature classification based on mathematical morphology method

Presents a new method used in fundus image processing and its artery-vein cross feature classifying based on mathematical morphology. Some of the selected morphological structuring elements and their combinations with several particular morphological operations are applied to the gray level image and the binary image. This paper also compares this method with convolution method and template matching method. The result shows that the morphological method is more effective. The gray level profile of the vein section along artery axis near the cross point is selected as the quantitative criterion to classify cross phenomena.