Chroma Prediction Linear Mapping with Reduced Reference Filtering

The patent extracts only the necessary reference samples from the luma neighbouring region instead of using all available samples. By selecting a subset of reference samples that are most relevant for prediction, the method reduces the number of blocks that need to be processed while maintaining prediction accuracy. This extraction approach directly addresses the contradiction by reducing computational complexity without sacrificing chroma prediction accuracy.

The patent segments the luma neighbouring region into distinct reference blocks and selectively processes only those blocks that contain valid reference samples. By dividing the neighbouring region into manageable segments and processing only the necessary portions, the method reduces overall computational complexity while maintaining the accuracy benefits of using multiple reference blocks for linear model construction.

The patent extracts and processes only the essential filtering operations on selected reference samples rather than applying filtering to all neighbouring blocks. By identifying and processing only the necessary samples that contribute to accurate prediction, the method reduces filtering operation workload while maintaining prediction accuracy. This selective extraction approach resolves the contradiction between accuracy and productivity.

The patent applies filtering operations to only a partial set of reference samples that are most critical for prediction accuracy, rather than excessively processing all available samples. This partial action approach ensures that sufficient filtering is applied to maintain accuracy while avoiding the excessive computational burden of processing every neighbouring block, thus resolving the productivity-accuracy contradiction.

The patent extracts and processes only the necessary reference samples from neighbouring blocks that are essential for achieving good chroma prediction quality. By removing unnecessary processing of redundant or less important blocks, the method maintains prediction quality while significantly improving video coding and decoding efficiency. This extraction principle directly addresses the efficiency-quality contradiction.

The patent segments the video coding process into distinct stages where reference sample selection and filtering are performed only on necessary blocks. By segmenting the processing workflow and applying down-sampling and linear model construction only where needed, the method maintains chroma prediction quality while reducing overall coding and decoding complexity, thus improving efficiency.