Investigation of filtering of rain streaks affected video sequences under various quantisation parameter in HEVC encoder using an enhanced V‐BM4D algorithm

Abstract

This study proposes an enhanced video block matching four-dimensional (V-BM4D) denoising algorithm for highly correlated rain streaks pattern video sequence in high-efficiency video coding (HEVC) encoder. Removal of structural rain streaks from high definition (HD) video sequences are challenging issues in real-time video application. Temporal information of original video sequences is distorted among the successive video frames due to the presence of natural degradation parameters such as fog, smog and rain streaks. Rain streaks are highly correlated structural noise pattern, which affects the nature of video frame compared to other natural degradations. Existing HEVC encoder is employed with in-loop filtering to protect the temporal information of original video sequences from rain streaks pattern. However, in-loop filtering is unable to remove the highly directional oriented rain pattern from video frames. To retain temporal information among the successive video frames, enhanced block matching denoising block is adopted in HEVC coder. The proposed enhanced VBM4D- based in-loop filtering eliminates the various level of rain streaks from the HD video sequences by temporal information prediction in 4D platform. Experimental results demonstrate that the proposed denoising algorithm yields better peak signal to noise ratio value and provides better bit rate saving for various HEVC configuration.