Inertial graphic gravitational random walk for network structure image segmentation

Abstract

Network structure image, such as retinal blood vessels, has many important applications in medicine, biometric identification and other fields. The traditional image segmentation methods for network structure images usually face the challenge that the region of interest (ROI) is broken. To tackle this challenge, this paper presents a mechanism of random walk walker movement based on the central gravity of ROI. The proposed approach exploits the gravity of the seed point in the walker's visual field, and the continuity of the ant movement path, to segment the network structure region without broken. Experimental results are presented to show the superior performance of the proposed approach against the conventional image segmentation approaches.