Application of SVM-based Fuzzy Inference System for Image Denoise

1 This work is supported by national natural science foundation of China, No.60475036. Abstract –The noise detection model is build up by using SVM-based fuzzy inference system for detecting the impulse noise in an image. The support vectors which are used to confirm the fuzzy basis function and create the corresponding fuzzy rules are extracted from the training samples by the learning mechanism of support vector machine (SVM). Accordingly, a noise detection system based on the fuzzy reasoning model is build up. This system is composed of two SVM-based fuzzy inference subsystems and a decision subsystem. Thereinto, the inference subsystems detect noise information on horizontal and vertical orientation, respectively; and the decision subsystems make decision by synthesizing horizontal and vertical information. The experiment results show that the proposed method can efficiently detect and remove noise while preserving the detail information.