Video Block Prediction Fusion for Cross-Component Coding

The patent merges multiple prediction methods by fusing the first prediction (from first CCP candidate) and the second prediction (from second CCP candidate) to generate a fused prediction for the current video block. This combination of multiple prediction sources improves both coding efficiency and effectiveness by leveraging complementary information from different prediction candidates.

The patent creates a composite prediction signal by combining multiple prediction candidates (first CCP candidate and second CCP candidate) into a fused prediction. This composite approach integrates strengths from different prediction sources, achieving superior coding performance compared to single prediction methods.

The patent applies partial fusion by selectively fusing predictions only for specific video blocks where multiple CCP candidates are available and beneficial. This approach improves coding effectiveness without unnecessarily increasing complexity for all blocks, achieving a balanced solution.

The fused prediction acts as an intermediary that combines information from multiple CCP candidates. This intermediate fused prediction then serves as the basis for encoding the current video block, simplifying the overall process while improving effectiveness compared to evaluating multiple separate predictions.