Skip to Main Content
A Takagi-Sugeno (TS)-type fuzzy system trained by support vector machine (TSFS-SVM) is proposed and applied to skin color segmentation. In TSFS-SVM, the consequence of each rule is of TS-type. Instead of being trained by neural learning, TSFS-SVM is trained by SVM with the objective to obtain a higher generalization ability. Performance of TSFS-SVM is verified through skin color segmentation problem. To represent color information by histogram as accurately as possible, non-uniform partition of HS space is used. Histogram information from images under different environments is used to train TSFS-SVM. Advantage of TSFS-SVM is verified by comparisons with other compared methods.