Abstract

High-definition (HD) video data imposes significant storage and bandwidth requirements, particularly in real-time applications such as video streaming and telemedicine. Wavelet-based video compression has emerged as a viable solution due to its multiresolution representation and scalability. Despite its advantages, traditional wavelet-based compression techniques suffer from spatial and temporal redundancy, especially for HD videos with complex motion dynamics. This redundancy leads to suboptimal compression efficiency and quality degradation. This paper proposes an improvised hybrid redundancy reduction framework that integrates motion-compensated temporal filtering (MCTF), adaptive lifting schemes (ALS), and directional intra-frame prediction (DIP) into the wavelet video codec. Additionally, a content-aware entropy coding module is introduced to adapt to varying motion intensities in HD sequences. The method includes a dynamic GOP (Group of Pictures) size selector based on scene complexity, further optimizing redundancy handling. The proposed method was benchmarked using standard HD video sequences (720p and 1080p) on MATLAB and compared with four existing hybrid methods: MC-EZBC, SPIHT-MCTF, H.264-DWT, and 3D-SPIHT. Our method achieved a PSNR improvement of 1.8–2.5 dB, bitrate reduction of 12–17%, and SSIM improvement of 0.025–0.045 on average. Subjective analysis also confirmed better perceptual quality, particularly in high-motion scenes.

Authors

A. Sevuga Pandian¹, Leti Wakuma²
Kristu Jayanti College, India¹, Dambi Dollo University, Ethiopia²

Keywords

Wavelet Compression, Redundancy Reduction, Motion-Compensated Filtering, HD Video Compression, Content-Aware Encoding