H.264 region-based weighted prediction for scenes with local brightness variations

This paper presents a novel region-based weighted prediction (WP) parameter estimation scheme for H.264 to encode scenes with local brightness variations (LBV). The scheme is achieved by utilizing multiple WP parameter sets for various regions and assigning them to the same reference frame. These multiple WP parameter sets are then encoded using the framework of multiple reference frame motion estimation (MRFME). This arrangement ensures that the proposed scheme is compliant with the H.264 standard. In this scheme, a region partitioning process is designed to divide the current frame into different regions where each one has some degree of uniformity in its brightness variation. This can assist in estimating multiple sets of region-based WP parameters accurately. As a consequence, the proposed scheme can improve prediction in scenes with different degrees of LBVs in different regions of the same picture. Results show that the region-based scheme can efficiently handle LBV scenes with significant coding gain.