In this paper, two prediction compensated multiple description coding algorithms are proposed based on H.264 intra frame coding and JPEG 2000, respectively. The image is split into two parts. Each forms the base layer of one description. The prediction residual of the other part is encoded in each description as the enhancement layer. In the H.264-based approach, we use the H.264 intra frame prediction to estimate the other description, and we investigate the performances of different partitions of the image. In the JPEG 2000-based approach, the least-squares algorithm is used for inter-description prediction. Image coding results show that performances of the proposed methods are comparable with that of the latest MDC results while the complexity is reduced.
展开▼
