A region-based video de-noising algorithm based on temporal and spatial correlations

Abstract

Video de-noising is important for display and subsequent analysis. However, it remains challenging with regard to structured noise and temporal smoothness. This paper proposes a region-based video de-noising algorithm. First, the luma component of each frame is explicitly divided into a moving region and a non-moving region. Then, 3D structures are formed for different regions according to their different temporal characteristics. A bilateral filter is extended to 3D and subsequently employed for luma component de-noising. Second, de-noising for chroma is proposed instead of a simple transplantation of the de-noising for luma. The edge information within the de-noised luma is referred to during chroma de-noising based on the joint bilateral filter. The experimental results demonstrate that the proposed de-noising algorithm, which outperforms state-of-the-art algorithms, can not only efficiently remove the noise but also maintain temporal smoothness among neighboring frames.