Improved Canny Edge Detection Algorithm

This paper introduces an improved Canny edge detection algorithm. The algorithm replaces traditional Gaussian filtering with bilateral filtering, which better preserves edge structures and ensures more precise edge detection in detailed regions. Additionally, the algorithm employs adaptive histogram equalization to enhance the local contrast of the image. At the same time, the adaptive thresholding in the algorithm enhances the detection performance for edges of varying intensities, such as weak and strong edges, particularly in low-contrast areas. During non-maximum suppression, a neighborhood comparison of edge pixels is manually implemented based on gradient direction to ensure that only local maximum edges are retained, increasing the precision in processing gradient directions, particularly for slanted or subtle edges. Finally, the algorithm applies morphological techniques to refine edges. Experimental findings show that the enhanced algorithm boosts the accuracy of object edge detection from 0.717 to 0.750.