Cross-Component Sample-Adaptive Offset Coding for Lower Bit Rates

The patent introduces luma samples as an intermediary to derive chroma sample offsets. Instead of directly encoding chroma blocks, the method uses luma sample characteristics (through classification and offset derivation) as a mediator to improve chroma prediction accuracy. This intermediary approach leverages the correlation between luma and chroma components to enhance coding efficiency without requiring complex chroma-specific processing.

The patent changes the parameter representation by deriving chroma offsets from luma sample values rather than using fixed or simple gradient-based offsets. The classification of luma samples into different categories (e.g., based on sample positions or value ranges) and the subsequent derivation of offsets based on these classifications represent a parameter transformation that adapts the prediction to local image characteristics, improving efficiency while maintaining manageable complexity.

The patent segments the chroma coding process into distinct stages: luma sample classification, offset derivation for different chroma components (Cb and Cr), and offset application. By dividing the processing into these manageable segments, the method reduces bit-rate requirements through efficient cross-component prediction while keeping each processing stage relatively simple and systematic.

The patent performs preliminary classification of luma samples and derivation of offset values before the actual chroma prediction and encoding. This preliminary action prepares the necessary parameters (classification results and offset values) in advance, allowing the main encoding process to proceed efficiently with reduced bit-rate requirements without requiring complex real-time processing during chroma encoding.