Binocular Stereo Vision Based On Biological Vision Principle

Basing on the recent researches on biovision, this paper makes a detailed research on the binocular biovision model.There are two channels in biovision system to extract boundary feature and brightness feature respectively. A detailed analysis on the dynamics of the two channels when they processing monocular image is presented. Integrating the researches on biovision, we rectify the mistakes in the original model and design a new elliptic sharpening template to enhance its effectiveness to the fuzzy boundaries as well as its efficiency.It's very complex for analyzing the dynamics of biovision model processing occlusion-free binocular images. This paper introduces some essential perception characteristics in binocular biovision and indicates that the models presented in recent years can't process the images of real-world scene. Consequently we designs a new adaptive disparity filter which implements the binocular matching of disparity features in stereo images, and constructs a binocular biovision model as well as a stereopsis system based on the disparity filter to gain depth perception of surfaces through binocular features. By using this stereopsis system, depth perception of surfaces in real-world stereo images is simulated and realized.When the dynamics of biovision model processing binocular images with all kinds of occlusion is analyzed, integrating the original models, we design a monocular feature filter basing neuron dynamics, develop further the adaptive disparity filter and propose a novel computational model of stereo vision that can gain the depth perception of monocular features. This model can realize the depth perception of the objects and the figure-ground separation not only in the normal scenic images but also in the occluded scenic images. Experimental results testified the model's feasibility.