Video Decoding Using Dynamic Reference Vector Selection

The patent dynamically determines reference vector types (motion vector or disparity vector) based on prediction mode and neighboring block characteristics, rather than using a fixed block-based approach. This dynamic adaptation allows the system to optimize coding rate by selecting the most appropriate reference vector type for each block while maintaining manageable encoding complexity through systematic decision rules.

The patent changes the parameter of reference vector type selection based on prediction mode and neighboring block information. By modifying how reference vectors are determined (using depth values, neighboring block types, and prediction modes), the system improves coding rate performance without requiring fundamentally complex encoding architecture.

The patent applies local quality by determining reference vector types based on specific neighboring block characteristics and prediction modes rather than uniformly processing all blocks. This localized approach improves coding efficiency by adapting to local block properties while avoiding the computational burden of analyzing all possible neighboring blocks for every encoding decision.

The patent uses partial action by selectively utilizing neighboring block information based on prediction mode and block type requirements. Rather than comprehensively analyzing all neighboring blocks, the system applies reference vector determination only where necessary and appropriate, improving coding efficiency without excessive computational complexity.