The New Approach Of Image Smoothing And Edge Detection Based On Non-linear Fuzzy Function

Image smoothing and edge detection are the basic operations during image processing.The most important and prevalent algorithm using in the smoothing and edge detectionoperations is the algorithm based on fuzzy idea. During the corresponding algorithm proposedin this paper, three fuzzy are concerned: neighborhood degree, low saturation and lowintensity, fuzzy edge. In this paper, first for the first two fuzzy concepts, we use non-linearfuzzy functions for color image smoothing and edge detection, which based on the linearfuzzy functions;After that, we propose mathematical morphological methodology to processthe last fuzzy concept—fuzzy edge;In the middle of them, the concept of removed-noisethreshold and new definition of image pixel's contrast is proposed.During the processing of image smoothing, this paper presents a new definition ofneighborhood degree based on Cauchy distributing fuzzy function instead of the traditionallinear neighborhood definition degree, which makes the location of the neighborhood degreemore precise;Then, this paper presents an efficient smoothing algorithm based onremoved-noise threshold, which causes the image more explicit and shows obvioussuperiority especially when the noise is more complex. Results show that the approach basedon fuzzy smoothing approach provides a powerful noise removal ability than the classicalones.During the processing of image edge detection, this paper presents a fuzzy functionbased on sin transform to locate the fuzzy sets of low saturation and low intensity. After that,we combine them with hue contrast and results in normalized hue contrast, which can getmore efficient information of the edge. The methodology of using mathematical morphologyto process the fuzzy edge is introduced in succession. The basic principle of the HSV vectorsorting is utilized during it. Results show that the image edge of using the fuzzy functionbased on sin transform is better than that of using the fuzzy function based on linear function.Meanwhile, the fuzzy edge is located well by mathematical morphology.