DXYW: a depth-guided multi-channel edge detection model.

The study of biologically inspired edge detection is one of the hot research topics in the computer vision field. Previous biologically inspired edge detection models are mainly based on X-type ganglion cells. However, according to physiological studies, Y-type and W-type ganglion cells exist in the visual system, playing essential roles in visual information processing. All three types of ganglion cells form their own independent visual channels, and all three channels working parallelly demonstrate a depth-guided stereoscopic mechanism. To model the biological visual mechanism more comprehensively, we propose a biologically inspired edge detection method based on the visual mechanism of the X-, Y-, and W-channels, namely DXYW. In the proposed model, three separate sub-models simulating the X-, Y-, and W-channels are designed to process grayscale images and disparity maps. Furthermore, a depth-guided multi-channel fusion method is introduced to integrate the visual response from different channels. Experimental results show that the multi-channel mechanism modeled in DXYW is consistent with the relevant physiological mechanisms. And DXYW achieves competitive edge detection performance that better maintains the integrity of object contours and suppresses background texture. [ABSTRACT FROM AUTHOR]