Adaptive optimization of non-constant luminance to constant luminance for HDR video distribution
Fujun Xie, Maryam Azimi, Ronan Boitard, Mahsa T. Pourazad, Panos Nasiopoulos · 2018 International Conference on Electronics, Information, and Communication (ICEIC) · 2018
Traditional pixel representations decompose luma and chroma information for improved compression efficiency. One widely-used example is the Y'CBCRcolor pixel representation which is mainly used in image and video compression. Two methods exist for converting RGB pixel representation to Y'CBCR: Non-Constant Luminance (NCL) and Constant Luminance (CL). CL equations are derived from the weighted combination of the targeted gamut color primaries in the light linear domain. NCL method applies the same equations on perceptually encoded values, thus leading to reduced compression efficiency and lower color quality compared to those of CL. However, given the increased implementation cost of CL, broadcasting companies and television manufacturers choose to use NCL. In this work, our motivation is to derive new equations that take advantages of the CL method's efficiency without increasing hardware complexity for High Dynamic Range video distribution. Our proposed method is designed to be content adaptive. Results indicate that improved color quality can be achieved in terms of DE100 metric for sequences with one prominent primaries (Red, Green, or Blue) when compared with the NCL approach.