Complexity efficient stopping criterion for LDPC based distributed video coding

In several distributed video coding architectures, a well-known complexity trade-off exists, where the low encoding benefits are paid with a higher decoding complexity. In a feedback channel based DVC architecture, the high decoding complexity is mainly due to the Slepian–Wolf decoding and the repetitive requestdecode operation, especially when there is no initial encoder rate estimation or iterative motion refinement is employed. In this paper, an early stopping criterion for the LDPC syndrome belief propagation decoder is proposed that is able to reduce the number of decoding iterations. As a consequence, a significant reduction of the DVC decoder complexity can be observed with negligible losses in RD performance. The experimental results show reductions up to about 4 times in decoding complexity with a maximum of 0.15dB loss at high bitrates while for low and medium bitrates the RD performance loss is negligible.