A Bionic Contour Detection Method Based on Dynamic Receptive Field

Abstract

The present invention aims to provide a bionic profile detection method based on a dynamic receptive field, comprising the following steps: A, input the image to be detected through gray scale processing, and calculate the spatial standard deviation of each pixel; B, combine each pixel's The spatial standard deviation calculates the optimal response value of the classic receptive field of each pixel; C, calculates the final contour response value of each pixel; D, calculates the final contour value of each pixel. The method overcomes the defects of the prior art, and has the characteristics of good bionic effect and high contour detection rate.

Description

technical field [0001] The invention relates to the field of image contour detection, in particular to a bionic contour detection method based on a dynamic receptive field. Background technique [0002] Contour detection is an important basis for technologies such as object detection, shape analysis, object recognition, and object tracking. In the vision system, the dynamic characteristic runs through the whole process of contour extraction. Due to the difference in local image information, when the visual system processes information, the size of the receptive field is also changing dynamically. At present, many contour detection models inspired by biology simply simulate part of the physiological characteristics of the visual system. In the process of contour extraction, some visual dynamic characteristics cannot be well simulated, resulting in the lack of contour information and the enhancement of texture information to a certain extent, so that the integrity of the targ...

AI Technical Summary

Problems solved by technology

Due to the difference in local image information, when the visual system processes information, the size of the receptive field is also changing dynamically. At present, many contour detection models inspired by biology simply simulate part of the physiological characteristics of the visual system. In the process of contour extraction, some visual dynamic characteristics cannot be well simulated, resulting in the lack of contour information and the enhancement of texture information to a certain extent, so that the integrity of the target contour cannot be well guaranteed.

Images