Method for detecting hard exudation in eye fundus image based on multi-light-source color constant model

Abstract

The invention discloses a method for detecting hard exudation in an eye fundus image based on a multi-light-source color constant model, belongs to the technical field of image processing, and solvesthe problem that the hard exudation detection precision is low due to the fact that a single light source is adopted for color correction in the prior art. The method comprises the following steps: inputting an original fundus image, preprocessing; performing color correction on the preprocessed original fundus image by using a multi-light-source color constancy algorithm; carrying out optic discpositioning by combining the blood vessel information in the fundus image subjected to color correction, and segmenting an optic disc area by utilizing rapid mean shift to obtain an optic disc image;performing threshold segmentation and morphological reconstruction on the basis of the optic disc image and the image obtained by preprocessing, and extracting a hard exuded candidate region to obtaina candidate region image; and extracting color histogram features, color constancy features and texture features of the candidate region image; and using the extracted features for detection to obtain a detection result. The method is used for feature extraction and hard exudation detection of the fundus image.

Description

technical field [0001] The invention discloses a method for detecting hard exudation in a fundus image based on a multi-light source color constancy model, which is used for feature extraction and hard exudation detection on the fundus image, and belongs to the technical field of image processing. Background technique [0002] In the existing technology, the fundus image is used to extract the area, circumference, roundness, diameter, average gradient and other shape features in the candidate area, or use a deep neural network to extract depth features for further subsequent hard exudation detection, but Based on the extracted features, the detection accuracy is not high. [0003] In the prior art, color correction technology is also used for hard exudation detection, but the color correction method assumes a single light source, and usually assumes that the spectral distribution of the source in the scene is relatively uniform. In a real environment, this assumption is dif...

AI Technical Summary

Problems solved by technology

[0004] In view of the above-mentioned research problems, the purpose of the present invention is to provide a method for detecting hard exudation in fundus images based on a multi-light source color constancy model, which solves the problem of subsequent har

Images