Application Of Visual Perception Process In Object Contour Detection

The outline describes the overall shape of the object,which is an important feature of human object recognition.In the field of computer vision,the perception of the object contour is a main research direction.In the complicated background of the natural image,the outline of the object is easily mixed with the background of the texture,and the contour detection is very difficult.How to remove a large number of false contour edges generated by the interference of the background texture,and realize the detection of the contour of the object,which is an urgent problem to be solved at present.The human visual perception system can process the visual information rapidly and perfectly,which has a good recognition ability to the external visual information.Therefore,the study of human visual perception mechanism,and then establish the model of visual perception processing information,not only is the key to solve the problem,but also one of the hot spots of domestic and foreign research.The main purpose of this paper is to analyze the relationship between the receptive field central neurons and their neighboring neurons,simulating visual suppression between nerve cells and enhance the effect.A contour detection model is proposed to meet the characteristics of visual perception,completing natural images in complex background contour detection.Firstly,summarizes the research results of visual cortex cell physiologist structure properties and function in recent years,and the classic contour detection model;Secondly,in combination with the non-classical receptive field,a modified contour detection model is established to overcome the shortcomings of Grigorescu et al;Subsequently,the mapping between the image information and the cell response was established according to the windmill arrangement characteristics of the visual cortex cells,and the enhancement of the contour and the suppression of the background texture were realized through the horizontal connection mechanism;Finally,because the existing algorithms still have self inhibition phenomenon,contour continuity is not good,put forward another contour enhancement algorithm.The algorithm is introduced the multilevel surround inhibition into the isotropic model,and puts forward the combination of the improved algorithm,then based on the Markov random field theory,the contour probability model is established to remove the edge of the noise and protect the weak contour.Solving the problem of complex background texture interference and weak contour detection in the contour detection,the continuity of the output contour is better,and the detection performance is higher.