Video Coding with Cross-Component Residual Models

The patent applies cross-component residual models (CCRMs) that establish statistical relationships between different color components (luma and chroma). By changing the parameter representation from independent component coding to correlated component coding, the patent improves coding efficiency. The CCRM uses parameters such as offset values and scaling factors to model the cross-component relationships, thereby reducing the residual energy that needs to be encoded and achieving higher coding gain.

The patent implements CCRM as an optional enhancement to conventional video coding standards. The encoder can selectively apply CCRM to specific video units or regions where cross-component correlation is beneficial, rather than applying it universally. This partial application approach achieves improved coding gain while controlling the increase in computational complexity, as the patent allows traditional coding paths to remain available when CCRM is not advantageous.