Research And Application On A Bionic Model Of Visual Nerve Systems

Human get to know the world mainly by our vision,hearing,smelling senses and so on,one of most important and effective way to get information for brain is vision.At the same time,visual nerve system is very complicated.According to the physiology research of visual system,the relationship between layers of retina,the neural connection of retina layer and lateral geniculate(LGN),LGN and visual cortex,and the transitive relationship of sublayers in primary visual cortex were analyzed and summarized in this paper.According to the physical anatomy structure of visual system,an integrated visual system model was presented based on bionics theory.The model has some characteristics as follows:(1)To scale,retina model is constructed based on neural ensemble to assure the homogeneity of mesoscopic scale of different parts in retina.(2)To structure,the model follows anatomy of retina and includes main parts of retina,such as retina model and primary visual cortex model.(3)To neuron,suitable neuron models are selected to simulate action of different neurons according to related research productions.Analysis results to proposed retina model show that the architecture of retina model has the characteristics of small-world network,which is coincident with related analysis results,at the same time,to the characteristics of dynamics,the model shows steady state similar to limit cycle to sustained and changeless input.Finally,the proposed model was used in salient image recognition.According to structure of the model,the study rule was made,and image was preprocessed by the method of histogram equalization and smooth filtering.Then the preprocessed image was input by the image processing function in the software of Matlab.The resutlts showed that the proposed model in this paper was superior to Itti model in the field of salient image recognition.