Multiple weighted prediction models for video coding with brightness variations

Abstract

Weighted prediction (WP) is a tool introduced in H.264 to improve motion-compensation performance for video sequences with brightness variations. Various WP models to estimate the parameter set have been discussed in the literature. However, no single WP model works well for all types of brightness variations. A single reference frame multiple WP models (SRefMWP) scheme is proposed to facilitate the use of multiple WP models in a single reference frame. The proposed scheme makes a new arrangement of the multiple frame buffers in multiple reference frame motion estimation. It enables different macroblocks in the same frame to use different WP models even when they are predicted from the same reference frame. Furthermore, a new reordering mechanism for SRefMWP is also proposed to guarantee that the list of the reference picture is in the best order for further decreasing the bit rate. To reduce the implementation cost, a reduction in the memory requirement is achieved via look-up tables (LUTs). Experimental results show that SRefMWP can achieve significant coding gain in scenes with different types of brightness variations. Furthermore, SRefMWP with LUTs can reduce the memory requirement by about 80% while keeping the same coding efficiency as SRefMWP.

Efficient temporal and interlayer parameter prediction for weighted prediction in scalable high efficiency video coding

“…If the proposed approaches cannot be applied, the conventional HWPP would be performed. In detail, in L B , if the reference frame is in list 1 , EDWPD first attempts to predict the WP parameter from each reference frame in list 0 based on (11). Otherwise, if there is no match or if the reference frame is in list 0 , we would check whether there are two reference frames with different POC so as to avoid TD D becoming zero in Eq.…”

“…Zhang and Cote 7 and Aoki and Miyamoto 8 proposed a more accurate WP parameter estimation, which uses the AC and DC characteristics. Tsang et al [9][10][11] proposed to have multiple WP parameters for one single frame to get higher coding efficiency. In Refs.…”