A Simple Effective Noise Smoothing Filter Using Grayscale Minimum Variance

Based on the filter using grayscale minimum variance, an improved algorithm for noise suppression is proposed in this paper. The approach first constructs nine kinds of filter windows surrounding the object pixel and lets them move together on the image canvas. Then it calculates the respective grayscale variance of the pixels under each window and selects a window area with minimum grayscale variance. Lastly, the method judges whether the gray value of the object pixel can be replaced by the average gray value of pixels within the selected pixel area based on their grayscale difference. The experimental results show the proposed method has good performance.