Generalized inter-layer residual prediction for scalable extension of HEVC

Scalable video coding extension of HEVC (SHVC) is being developed by Joint Collaborative Team on Video Coding (JCT-VC) of ISO/IEC MPEG and ITU-T VCEG. Different from scalable video coding extension of H.264/AVC (SVC), SHVC employs a multi-loop decoding framework so that the inter-layer residual prediction in SVC does not perform well in SHVC. In this paper, a method called generalized inter-layer residual prediction (GILRP) is proposed. To improve prediction accuracy, the residual predictor is derived with the information from both base and enhancement layers. Moreover, three additional weighting types are introduced on top of inter coding modes to further compensate errors caused by base layer quantization. Simulations under SHVC common test conditions defined by JCT-VC show that 2.9%, 5.1% and 4.9% overall luma BD-rate reduction on average were obtained over SHVC reference software for configurations of random access, low delay with P slices, and low delay with B slices, respectively.