A contour detection method based on non-classical receptive field subfield
Zhang Jingyan
Fan Yingle
Fang Tao
Abstract：Objective This paper proposes a novel contour detection method inspired by the surround inhibition mechanism of the primary visual cortex.Methods The method involves simulating the response characteristics of the classical receptive field in the primary visual cortex to external stimuli and constructing a multi-directional two-dimensional Gabor filter model for extracting primary contours.A non-classical receptive subfield surround suppression model is proposed based on the structural characteristics of the non-classical receptive subfield for texture suppression.Additionally,a two-dimensional Gaussian function is used to simulate information processing by ganglion cells,and information is transmitted across levels to improve the response rate.Finally,the characteristics of capturing global information by the human eye are simulated to correct the contours and obtain the final contour map.Results Qualitative and quantitative analysis compared with other existing contour detection algorithms;The average accuracy(AP)of any 200 images in the BSDS500 reached 0.703.Conclusion the results show that the proposed algorithm can more effectively highlight the contour of the subject and suppress the texture background.
Keywords：contour detectionnon-classical receptive subfiledvisual pathwaysurround inhibition
Publication Date：2025-02-14
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)