Edge detection in time variant scenarios based on a novel perceptual method and a Gestalt spiking cortical model

Abstract

Based on recently neurocomputational models inspired on neural synchronization for perceptual grouping, we propose in this paper the Gestalt Spiking Cortical Model (GSCM) and the Perceptual Grouping segmentation (PGSeg). The GSCM is a network based on the mechanisms of perceptual grouping models designed to detect scene attributes with excitatory and inhibitory inputs. PGSeg is a neuroinspired method designed to detect object edges presented in video sequences that involve time variant scenarios. Experimental results using videos from the perceptual computing and ChaDet2014 databases, show that PGSeg has better performance regarding edge detection and edge coherence through video sequences.