摘要：In order to suppress image noise effectively and extract the edge clearly and accurately, a novel approach based on omni-directional and multi-scale mathematical morphology for edge detection is proposed in this paper. In order to compare different object geometries a modified Hausdorff distance application is composed. With a new moments rule used, Opening operation and closing operation of large scale structural elements are adopted for suppression of noise, while small scale structural elements are used for edge extraction. Finally, the weighting combination calculation of edge maps acquired under different directions is executed according to different weights. The experimental results show that the improved algorithm has a great promotion in extracting edge information compared with several reference algorithms.
关键词：omni-directional;multi-scale;Hausdorff;edge detection;morphology;structural elements
