Videodisc positioning method combining Gbvs model and phase consistency

Abstract

The invention discloses a videodisc positioning method combining a Gbvs model and a phase consistency. The method comprises the following steps of (1) extracting a green channel of a color eye fundus image; (2) calculating a mask image of the fundus image and acquiring a region of interest (ROI) of the fundus image; (3) extracting three characteristics of brightness, a brightness contrast and the phase consistency; (4) using an improved significant model based on graph (graph based visual saliency, Gbvs) to construct a saliency graph; (5) extracting eye fundus image vein blood vessel skeleton lines and using a least square method to carry out parabola fitting; and (6) comparing significance in a parabola summit neighborhood and an average of significance of a whole eye fundus image and determining a videodisc position. During a process of calculating the saliency graph, according to a characteristic of a videodisc in the eye fundus image, the brightness, the brightness contrast and the phase consistency are extracted, the significant model based on the graph is improved, and during a process of positioning the videodisc, a structure characteristic of the vein blood vessel is used to assist accurate positioning. An experiment proves that the method possesses a good effect.

Description

technical field [0001] The invention belongs to the technical field of image processing, and relates to an optical disk positioning method combined with a Gbvs model and phase consistency. It can be used to locate the optic disc on fundus images with different degrees of normal and lesion. Accurate optic disc positioning can assist in locating the position of important tissue structures in the retinal fundus, and plays an important role in fundus image analysis. Background technique [0002] The processing and analysis of retinal fundus images can help doctors to screen and diagnose eye diseases and systemic diseases such as diabetes, and track the development of diseases. In the color retinal fundus image, the optic disc is a sub-circular, nearly yellow or white bright spot, accounting for about one-seventh of the fundus image in diameter. In addition, since the optic disc is the origin of the optic nerve and the origin of fundus blood vessels, there are a large number of...

AI Technical Summary

Problems solved by technology

Analyzing the existing optic disc localization methods in fundus images, the optic disc localization method using the appearance characteristics of the optic disc is suitable for normal fundus images, but when there are fundus lesions with similar brightness to the optic disc, it is easily disturbed by lesions and has poor robustness ; and for the optic disc positioning method based on blood vessel segmentation, because the blood vessel segmentation itself is a complicated task, the optic disc positioning based on this method often takes a long time and cannot meet the real-time requirements of fundus optic disc detection

Images