Toward a More Robust Canny for Edge Detection
Zhen Liu, Mingzhe Liu, Xin Sunny Huang · 2022 5th International Conference on Pattern Recognition and Artificial Intelligence (PRAI) · 2022
From the first appearance of Canny to present day, few edge detection algorithms can be as critical as it is. Specifically, these non-Canny framework-based algorithms do boost some disadvantaged metrics under certain conditions, but the outstanding overall performance of Canny makes it still being universally applied. Actually, some improvements in disadvantaged metrics can be achieved directly within the framework of Canny. This has rarely been discussed in the past for the deficiency of the comprehensive analysis for Canny. So, this paper presents a more robust Canny that substantiates the idea, in which the amplitude were calculated using the fitting plane, and the threshold filtering was achieved with the standard deviation