An Image Filtering Model Based On Content And Rough Compatibility Relation
Zhang Xuemei, Zhu Jinjun · 2007
This paper introduces an image content filtering model based on Clustering and Rough Compatibility Relation. In order to reduce the influence by lighting, a new Gamma correction method is used to adjust the pixel values. This filtering model use a TLHS color model that improve on the saturation arithmetic based on GLHS color model. TLHS color model has a steady skin color clustering with great efficiency. And use a new image segmentation method combines FCM clustering arithmetic and rough set theory. At first, an information table reflecting the relation of space segmentation is constructed according to the FCM clustering result under different clustering numbers. Secondly the difference degree between objects is confirmed by value reduction and getting the power value of attribute. At last, the α consistent relation under the rough set is defined and the image is divided into segments.