期刊名称：International Journal of Signal Processing, Image Processing and Pattern Recognition
印刷版ISSN：2005-4254
出版年度：2014
卷号：7
期号：5
页码：309-322
DOI：10.14257/ijsip.2014.7.5.27
出版社：SERSC
摘要：In morphology, the structural elements of different sizes and shapes have different abilities in maintaining image detail and anti-noise, an adaptive algorithm of edge detection based on multi-structure and multi-scale element is presented in this paper. First, the existing morphological operators for edge detection are improved. Then, the weight coefficients of different scales and different shapes are determined respectively by using the method of morphological difference and calculating the information entropy of the edge-detected. Last, fuse together the edges to achieve the image edge. For the image with noise, the algorithm can filter out noise effectively, and the edge is continuous, smooth, clear outline. The experiment results show that the algorithm has good effect in edge detection, strong anti-noise capacity, and the objective evaluation and visual effect are good, too.
