Cross-Component Prediction Candidate Selection for Video Coding

The patent applies parameter changes by dynamically adjusting CCP model parameters (alpha and beta) based on local video block characteristics. The method derives multiple candidate CCP models with different parameter sets and selects the optimal model for each block, rather than using fixed parameters across the entire video stream. This local adaptation of parameters improves both coding efficiency and effectiveness.

The invention implements dynamics by creating a dynamic CCP model selection mechanism. Instead of using a static prediction model, the system maintains a list of candidate CCP models and dynamically selects the most appropriate model for each video block based on local characteristics. The model list is updated and maintained adaptively, allowing the coding system to respond to changing video content properties.

The patent applies segmentation by dividing the video content into multiple video blocks and applying different CCP models to different blocks. Instead of using a single complex model for the entire video, the method segments the prediction task into multiple simpler block-level predictions, each using an optimized candidate model. This reduces overall computational complexity while maintaining high prediction accuracy.

The invention implements partial action by deriving only the necessary number of candidate CCP models (typically 2-3 candidates) rather than exhaustively searching all possible parameter combinations. The method performs partial optimization by selecting from a limited set of candidate models derived from local video characteristics, achieving sufficient prediction accuracy without exhaustive computation.