With Version 2 of the high-efficiency video coding standard, a new compression efficiency tool targeting redundancies among color components is specified for all 4:4:4 profiles, and referred to as cross-component prediction (CCP). This paper describes and analyses two additional extensions to the specified CCP variant. In the first extension, an additional predictor is introduced. Particularly, beside the luma component, also the first chroma component can serve as a reference for prediction of the second chroma component. The second extension proposes a method for predicting the CCP model parameter from the statistics of already reconstructed neighboring blocks. A performance analysis of coding RGB content in different color representations is given in comparison with CCP and both extensions. Experimental results show that the proposed extensions can improve the compression efficiency effectively compared with CCP, when applied in the YCbCr domain.
