Wavelet Multiscale Products Based Genetic Fuzzy Clustering for Image Edge Detection Analysis

A new edge detection algorithm by combining multiscale wavelet technique and genetic fuzzy clustering algoithm is proposed in this paper (called WGFCA), which can realize edge detection of input image automatically. Based on the theory that signals and noise have different characters along wavelet decomposition scales, WGFCA firstly multiply the wavelet coefficient of input image at adjacent scales to enhance edge structure and suppress noise, then, in order to restrain noise further, WGFCA apply fuzzy median filter to the result obtained above. Finally, edge map of input image is obtained by the great unsupervised classifying technique-genetic fuzzy clustering based on an effective feature extraction algorithm. Experiment results demonstrated promising performance of the proposed edge detection algorithm.