Adaptive ε-filter based on signal-noise decorrelation

This paper presents an adaptive ϵ-filter based on signal-noise decorrelation. ϵ-filter is a nonlinear filter to reduce noise and can be applied not only to speech signals but also to image signals. The filter design is simple and it can effectively reduce noise with an appropriate filter parameter. However, it requires the parameter setting to reduce the noise effectively, and the parameter settings are empirically executed. To solve the problems, we employ the criterion based on signal-noise decorrelation and propose a method to optimize the parameter of ϵ-filter automatically. The experimental results are also given to show the effectiveness of the proposed method.