To search, Click below search items.


All Published Papers Search Service


A Multi-layers Neural Network Model based on Characteristics of Ganglions’ Receptive Fields in Retina and an Algorithm for Watchfulness-keeping


Wei Hui, Yang Xian-Bo


Vol. 6  No. 10  pp. 10-20


Machine vision is an active branch of Artificial Intelligence. An important problem in this area is the balance among efficiency, accuracy and huge computing. The visual system of human can keep watchfulness to the perimeter of visual field while at same time their central attention is focused to the center of visual field for fine information processing. This mechanism of computing resource assignment could ease the demand for huge and complex hardware structure. Therefore designing computer model based on biological visual mechanism is an effective approach to resolve problems in machine vision. In this paper a multi-layers neural model was designed based on the features of receptive field of ganglion in retina to simulate multi-scale perceptive fields of ganglion cell. The neural model could maintain the alert to the outer area of whole image while it was capturing and processing more important information locating at the image’s central part. It may provide valuable inspiration for the implementation of real-time processing and the avoidance of huge computing in machine vision.


Computer vision, Neuro-vision