A New Approach to Video Coding Using 3D Directional Filter Bank

AbstractThe high resolution 3-D and higher dimensional volumetric data are increasingly used in a wide variety of applications including biomedical imaging, video processing and compression. Multidimensional filter banks and its associated constructions provide a framework and serve as an efficient computational tool in the formation, representation, and processing of multidimensional data sets. This paper focuses on the construction of 3-D directional filter banks (DFB) and its signal representations for scalable video coding. In this paper we consider five such DFBs and analyse the performance to find the best suitable one for scalable video coding. The five different filters includes: Maxflat filter, Vk filter, kos filter, Sinc filter and Haar filter. The performance evaluation has been done which quantifies the factors like Peak Signal to Noise Ratio (PSNR) and Mean Square Error (MSE). In order to achieve better compression, a comparative study between RLC and Huffman coding schemes has been performed and the average codeword length is measured and tabulated. Among the five filters and two compression scheme it is observed that Kos filters with Huffman coding provide better compression and low Average codeword length