Multiple descriptions coding for H.264/AVC using coinciding A2 lattice vector quantizer

Applications involving multimedia communications are widespread. However available networks do not meet the users needs such as unlimited bandwidth and reliability. Therefore video compression techniques are used to decrease the size of the video data and error resiliency techniques are used to combat against channel failures. Multiple Descriptions Lattice Vector Quantization (MDLVQ) is a technique that combines these techniques and suits for robust data transmission over unreliable network channels. Multiple Descriptions Coinciding Lattice Vector Quantization (MDCLVQ) is a new MDLVQ scheme based on the coinciding sublattices of A₂ lattice. In this paper MDCLVQ has been employed in order to form an MD coding scheme for H.264/AVC video coding standard to increase the robustness of video transmission over error-prone communication channels. The proposed MD video coding scheme is applied to several reference video sequences. The experimental results show that the encoding performance of the scheme is increased while the average PSNR of the central decoder remains above 33.84. It means that error resiliency of the scheme is increased without significant drop in the reconstruction quality. In addition the proposed scheme is compared with renowned techniques. It is observed that the proposed scheme outperforms all the other algorithms in low bit rate regime.