2-D Block Kalman Filter with Colored Driving Source for Image Restoration Using Pixel Classification

Abstract

We present a fast image restoration method using the 2-D block Kalman filter with colored driving source. The proposed method has the remarkable feature that it can reduce the computational complexity using the high correlation pixels of current region and reference region, while the conventional 2-D block Kalman filter methods have been performed without considering the relationship between the reference region and the current region. We have shown the effectiveness of the proposed method by numerical results and subjective evaluation results.