System and method of implementing multiple prediction models for local illumination compensation

Abstract

A method is provided for inter-coding video in which coding units can be encoded using multiple local illumination compensation (LIC) values to more accurately and efficiently transmit and render video. In the method, two or more LIC values can be established for a single coding unit (CU) such that the CU can be coded using multiple LIC values instead of a single LIC value as employed in present systems.

1 . A method of decoding a bitstream comprising:
 (a) receiving a plurality of rectangular coding units at least one of which having information associated therewith related to at least one local illumination compensation model;   (b) selectively enabling or disabling for each of said plurality of rectangular coding units whether said at least one local illumination compensation model is used for decoding each of said plurality of rectangular coding units;   (c) when selectively said disabled for one of said plurality of rectangular coding units decoding said one of said plurality of rectangular coding units without reference to any of said at least one local illumination compensation model;   (d) when selectively said enabled for one of said plurality of rectangular coding units decoding said one of said plurality of rectangular coding units with reference to said at least one local illumination compensation model, wherein said at least one local illumination compensation model is a linear model, wherein said at least one local illumination compensation model includes a variable scaling factor, wherein said at least one local illumination compensation model includes a variable offset, wherein said offset and said variable scaling factor are based upon a least square error technique using subsampled neighboring pixels to said one of said plurality of rectangular coding units, and wherein said offset and said variable scaling factor are based upon pixels identified by motion vectors of said one of said plurality of rectangular coding units in a reference picture.   
     
     
         2 . A method of decoding a bitstream comprising:
 (a) receiving a plurality of rectangular coding units at least one of which having information associated therewith related to at least one local illumination compensation model;   (b) selectively enabling or disabling for each of said plurality of rectangular coding units, which are coded using an inter coding mode, whether said at least one local illumination compensation model is used for decoding each of said plurality of rectangular coding units;   (c) when selectively said disabled for one of said plurality of rectangular coding units decoding said one of said plurality of rectangular coding units without reference to any of said at least one local illumination compensation model;   (d) when selectively said enabled for one of said plurality of rectangular coding units decoding said one of said plurality of rectangular coding units with reference to said at least one local illumination compensation model, wherein said at least one local illumination compensation model is a linear model, wherein said at least one local illumination compensation model includes a variable scaling factor, wherein said at least one local illumination compensation model includes a variable offset, wherein said offset and said variable scaling factor are derived based upon a least square error technique using subsampled neighboring pixels to said one of said plurality of rectangular coding units, and wherein said offset and said variable scaling factor are based upon pixels identified by motion vectors of said one of said plurality of rectangular coding units in a reference picture.   
     
     
         3 . The decoding method of  claim 2  further comprising a plurality of local illumination compensation models for a corresponding plurality of said rectangular coding units. 
     
     
         4 . The decoding method of  claim 2  wherein said offset and said variable scaling factor are jointly determined based upon said least square error technique. 
     
     
         5 . The decoding method of  claim 2  further comprising