Bionic contour detection method based on color and gray feature fusion

Abstract

The invention aims at providing a bionic contour detection method based on color and gray feature fusion. The bionic contour detection method comprises the following steps of: A, inputting a to-be-detected image, carrying out the calculation of color features, extracting color components in the to-be-detected image, and carrying out filtering to obtain a filtering response value of each color component; B, calculating single antagonism response based on color features of each pixel point; C, calculating a color feature contour response value and a color feature optimal direction of each pixelpoint; D, performing gray processing on the to-be-detected image, and calculating gray features, namely, calculating a gray feature initial response value of each pixel point; E, calculating a classical optimal response value and an optimal gray feature direction of the gray feature of each pixel point; F, calculating a gray feature contour response value of each pixel point; G, calculating a final contour response value of each pixel point; H, and calculating a final contour value of each pixel point. The bionic contour detection method overcomes the defects in the prior art, and has the advantages of comprehensive operation feature and high contour recognition 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 fusion of color and grayscale features. Background technique [0002] In the field of image contour detection, bio-inspired contour detection models are currently one of the mainstream research directions. The bionic models that have been proposed generally consider the color information or grayscale information of the image alone, and do not consider the effective combination of the color information and grayscale information of the image, but simply use the single and double antagonism principle of color or the grayscale The central excitation-peripheral inhibition mechanism simulates the complex visual perception system, which to some extent causes the loss of target contour information, thereby reducing the performance of the contour detection model. Contents of the invention [0003] The present invention aims to provide a ...

AI Technical Summary

Problems solved by technology

The bionic models that have been proposed generally consider the color information or grayscale information of the image alone, and do not consider the effective combination of the color information and grayscale information of the image, but simply use the single and double antagonism principle of color or the grayscale The central excitation-peripheral inhibition mechanism simulates a complex visual perception system, which to some extent causes the loss of target contour information, thereby reducing the performance of the contour detection model

Images