模糊集理论在图像分割中的简单应用

In this paper, by improving the extraction of Roberts to the edge of the edge pixels and noise from the one-dimensional histogram removed, and then the rest of the pixel cluster, calculated the optimal partition threshold, and then set up goals and background pixels pixels of fuzzy sets, respectively Each pixel obtained under the goals and background pixels fuzzy set of fuzzy membership. According to the re-gray background and objectives of the distribution, respectively, choosing the right confidence interval, CI is not the point of pixels and then use its neighborhood pixels membership classification, to be the ultimate outcome of the division. In this paper, through the neighborhood information to restore the original point of noise information in order to effectively avoid the binary processing, filtering caused varying degrees of fuzzy images as well as the loss of image detail, so as to improve the overall image segmentation . Such methods will be effective two-dimensional histogram of the ideological split into one-dimensional histogram processing, makes a clear division to speed up the pace.