Lossy Video Compression Technique for High Quality Videos Using 3D-Biorthogonal Wavelet Transform

This paper presents a completely new range-forward 3D video compression algorithm based on the combination of 3D Biorthogonal Wavelet Transform (3D-BWT), scalar quantization and Huffman coding allowing compression and decompression of video with high quality. Spatial and temporal correlations are captured through the application of multi-resolution representations which are derived by the 3D-BWT in video data decomposition. This is followed by the application of scalar quantization that reduces the precision in which transformed coefficients were obtained, this results into extreme compression while quality degradation is controlled at an acceptable level. The quantization approach is best achieved using the Huffman coding scheme. The encoded coefficients optimize the bitstream and are well suited for transmission or storage. The transmission or storage of encoded coefficients is optimized using Huffman coding. The process of 3D-BWT inverse is then used alongside the dequantization process and Huffman decoding for video decompression. The proposed technique has been demonstrated through experimental results to improve on existing techniques with respect to a number of quality metrics including compression ratio, mean squared error, and peak signal-to-noise ratio. The evaluation confirms that compared to other approaches, the proposed approach performs better by achieving improved overall video performance whilst its efficiency in compression is high making it applicable for 3D video compression applications.