Multiple Description Video Cod Based On 3D-DCT

The need of transmission of video via IP networks becomes more and more. In order to achieve this goal, two problems should be solved. Firstly, the coding method should be simple and high efficiency. Simple method can be implemented by hardware easily, and requires lower power. High efficiency can decrease network bandwidth needed for communication. Secondly, the coding is robust, i.e, it is fit for transmission via IP networks. Even when some packets are lost during transmission, acceptable quality of video can still be reconstructed.To solve the above-mentioned two problems, this paper proposes a multiple description video coding method based on 3D-DCT and MDSQ. The system consists of several parts: video volume formation, 3D-DCT, MDSQ, RLE coding and entropy coding and so on. 3D-DCT is a symmetrical transform so the encoder and the decoder has similar computational complexity. There is no motion compensation in the proposed scheme, so the method is simple and can be easily implemented in hardware. Through the optimization of the quantization volume and scanning order, the 3D-DCT based video coding scheme can obtain satisfactory results both in compression ratio and in PSNR. Multiple description coding divides the source information into several descriptions, each description could reconstruct acceptable source information, and reduced the loss caused by the lost of some descriptions, and improve the reliability of the video coding method.Experiment results show that the 3D-DCT based multiple description coding scheme has a finer robustness while being simple and highly efficient. The scheme is suitable for video transmission on portable terminal.