Multi-Layer and Adaptive Edge Detection Method Based on Multi-Scale Gabor Wavelets

Abstract

Recent years, the common adaptive edge detection algorithms use merely the global character of images, and the local character is ignored. The lack of image information always leads these algorithms are difficult to detect the low contrast edges and sensitive to asymmetric illumination. A novel adaptive edge detection algorithm is proposed to improve the edge detection performance. First, the original image is filtered to obtain the edge response image by introducing multi-scale Gabor filters. Second, non-maxima suppression technique is used on the edge response image to get the coarse edge points. Then, the edge response image is delaminated to adaptively select different thresholds towards different pixels of different layers according to pixels local mean and variance values. The fine edge points are got after filtering the edge response image with these thresholds. Finally, combine the coarse edge points and fine edge points together and get the final edge. Experimental results show that the proposed algorithm achieves lower contrast edge detection and robustness to asymmetric illumination, has better edge detection performance.