Color image edge detection method based on fuzzy theory

Color Image Edge Detection Based on Gaussian and Triangular Membership Function Fuzzy Inference

This technology is a method for detecting edges in color images using fuzzy theory. It fuzzifies pixels using horizontal and vertical Gaussian membership functions, synthesizes the results for each direction to infer edge values, and then performs defuzzification filtering using triangular and Gaussian membership functions.

Edge detection is a core preprocessing step in image analysis, pattern recognition, and computer vision. However, the standard Canny edge detection method requires multi-stage processing, including smoothing, gradient estimation, non-maximum suppression, and double thresholding. Such methods struggle to flexibly handle image uncertainty, limiting their performance in detecting edges in color images.

This technology divides the grayscale range into three parts, applying triangular membership functions to two regions and a Gaussian membership function to the third during defuzzification to precisely adjust edge intensity by region. It can be applied to machine vision inspection, object recognition, and image segmentation preprocessing modules, allowing for more accurate contour extraction without multi-stage computation.

‍

‍

Key Features:

• A fuzzification step that maps each pixel's data in the input image to a value between 0 and 1 using horizontal and vertical Gaussian membership functions
• A fuzzy inference step that synthesizes horizontal and vertical edge detection data to infer the edge value of the pixel
• A defuzzification step that filters the image with inferred edge values using two triangular membership functions and one Gaussian membership function
• A step of filtering the first and second grayscale regions with their respective triangular membership functions and the third grayscale region with a Gaussian membership function

‍