| Hi-index | 0.01 |
Reversible variable length coding (RVLC) is an important robust entropy coding technique for error resilient video transmission. The popular DCT-based video codecs all can use it. RVLC supports both the forward and backward decoding, so that the error-free data in the backward direction can be recovered. But RVLC can not retrieve the intact data between separate errors. This will degrade its error resilience performance especially in high error rates. On the other hand, as context modeling is widely adopted in modern entropy coders, RVLC shows low coding efficiency. Therefore, our design aims to improve the traditional RVLC in both error resilience and coding efficiency.