Adaptive Scan Path Selection for Spatial Domain Video Coding

The patent applies preliminary action by pre-defining multiple fixed scan path patterns (e.g., zigzag, horizontal, vertical, diagonal scans) before encoding. The encoder selects the most appropriate pre-defined pattern based on block characteristics such as gradient magnitude or variance, rather than adaptively generating scan paths based on prediction errors. This prevents error propagation because the scan path selection depends only on the current block's statistics, not on previously decoded data, thus maintaining coding efficiency while ensuring reliability.

The patent applies dynamics by making the scan path selection adaptive to the characteristics of each individual block. Although the scan paths themselves are fixed and pre-defined, the selection of which scan path to use varies dynamically based on block-specific metrics such as gradient magnitude, variance, or energy distribution. This allows the system to maintain reliability through fixed scan paths while achieving improved coding efficiency through context-adaptive selection, resolving the contradiction between fixed reliability and adaptive efficiency.

The patent applies parameter changes by using simple, easily computable block characteristics (such as gradient magnitude, variance, or energy distribution) as the basis for scan path selection. These parameters can be calculated with minimal computational overhead during the encoding process. By changing the scan path pattern based on these simple parameters rather than complex adaptive algorithms, the system achieves improved entropy encoding performance while keeping the implementation complexity low and suitable for real-time video coding applications.