Morphological segmentation-based eye fundus image lesion detection method

Abstract

A fundus image lesion detection method based on morphological segmentation. First, the fundus image is smoothed and filtered, and the region growing method is used to locate the region where the optic disc is located; then the fundus background image with the exudate area removed is obtained through morphological processing, and is segmented by thresholding A mixed area image including blood vessels and hemorrhage is obtained; finally, the edge detection is performed by Kirsch operator and the blood vessel image is obtained by region growing. The invention avoids the possible interference caused by blood vessels with similar gray levels in other bleeding detection algorithms, and also speeds up the detection speed.

Description

technical field [0001] The present invention relates to a technique in the field of image processing, in particular to a detection method for hard exudation based on background restoration of a healthy region of a fundus image and a method for detection of hemorrhage based on edge detection and region growth. Background technique [0002] With the development of computer graphics processing technology, the analysis of fundus images no longer relies solely on ophthalmologists' naked eyes. out and bleed techniques, and a great deal of research has been done. Traditional manual qualitative analysis lacks quantitative means. Relying on the computer to quickly and reliably automatically identify the lesions in the fundus image can prevent doctors from injecting special drugs into patients to make the fundus image clear, and at the same time avoid the doctor's manual judgment of film reading, saving a lot of manpower, material resources and time. The implementation of lesion det...

AI Technical Summary

Problems solved by technology

[0016] In view of the inaccurate response to some weak edges in the prior art, the detection of hemorrhage areas connected to blood vessels is not ideal, and the calculation amount is large, and the calculation speed is insufficient, the present invention proposes a fundus based on morphological segmentation The image lesion detection method uses the gradient information, gray level information of the image and the characteristics of similar gray levels of blood vessels, through the exudation detection algorithm and the hemorrhage detection algorithm, and uses morphological means to extract features, eliminating the need for complex methods such as classification. It improves the real-time operation without losing accuracy. While adapting to different types of fundus images, it avoids the interference that blood vessels with similar gray levels may cause in other hemorrhage detection algorithms, and also speeds up the detection speed.

Images