Research Of X,Y,W Cell Visual Neural Mechanism In The Contour Detection

Human vision system has the functions of quickly extracting object contours,distinguishing colors,shape perception,orientation perception,etc.Therefore,bionic contour detection has become one of the research hotspots in the field of computer vision.The existing bionic contour detection models mainly focus on the linear characteristics of X-type cells and the nonlinear characteristics of Y-type cells,ignoring the eye movement mechanism of W-type cells.The eye movement mechanism of W-cells has the ability to highlight the target contour,and it can regulate the visual information response of X-cells by studying the eye movement mechanism of W-cells,and the fusion of visual channel information of X-type cells,Y-type cells and W-type cells.It can enhance the integrity and continuity of contours,which plays an important role in improving the performance of contour detection.By studying the influence of fixation fretting characteristics of W-type cell eye movement mechanism on the visual information feature extraction of X-type cells.A contour detection model is proposed to modulate the information response of type X cells in the primary visual cortex based on the fixed micromovement characteristics of W-type cells.The classical receptive field of modulated X-type cells will be simulated by a Gaussian first-order derivative filter in which the fixation center is located at the origin and the fixation center is shifted to multiple directions.Because the amplitude of fixation micromotion of W-type cells is related to the surrounding environment,the offset of fixation center is calculated from the local information of gray image.The response energy map modulated by W-cell fixed vision fretting will obtain the optimal energy map through the competition mechanism.Finally,Gaussian difference function is used to suppress the texture of the optimal energy map and obtain the final target contour output.In view of the problem of extracting feature information insufficiency in the existing single-channel detection model inspired by X-type cells or Y-cell response characteristics,this paper will study the solid-visual micro-motion characteristics of W-type cells formed,And combined with the feedforward transmission mechanism of parallel information response in the X,Y-cell visual pathway,A contour detection model is proposed,which combines the spatial summation characteristics of receptive fields of X-type and Y-type cells and the fixation fretting characteristics of W-type cells.Neurophysiological studies show that X-type cells,Y-type cells and W-type cells all perform visual processing with independent and parallel channels and integrate information in the V1 area of the primary visual cortex.Inspired by this physiological mechanism,this paper is modeled by respective physiological characteristics of X,Y and W type cells.And process the grayscale image and disparity image at the same time.Finally,the contour features of X,Y and W channels are effectively fused according to the fixation surface determined by the disparity image.The qualitative and quantitative experimental results of Berkeley Segmentation Data Set(BSDS)and New York University Indoor Scene Dataset(NYUD)show that the proposed detection model can better retain the contour integrity and suppress the background texture,and obtain more continuous contour features.In the NYUD data set,the ODS is 0.65,OIS is 0.67,and AP is 0.63,so the contour detection model has high robustness.