Adaptive GOP Sizing Based on Motion Complexity in Video Encoding

The patent applies dynamics by transitioning from a fixed GOP size to a dynamically adaptive GOP size that changes based on motion complexity. The encoder calculates motion complexity metrics (such as motion vector magnitude, motion vector difference, or prediction error) for different candidate GOP sizes and selects the optimal size adaptively. This allows the encoding structure to dynamically adjust to content characteristics, resolving the contradiction between operational simplicity and encoding performance.

The patent changes the parameter of GOP size from a fixed constant to a variable parameter that adapts based on motion complexity. By computing motion complexity for multiple candidate GOP sizes and selecting the optimal one, the system transforms the GOP size parameter to respond to content characteristics, thereby improving rate distortion performance while maintaining encoding feasibility through automated selection.

The patent segments the encoding process into distinct stages: motion complexity calculation for multiple candidate GOP sizes, evaluation of rate distortion metrics for each candidate, and selection of the optimal GOP size. This segmentation allows the complex adaptive process to be broken down into manageable computational steps, making the increased device complexity more acceptable by organizing the computation efficiently.

The patent computes motion complexity for multiple candidate GOP sizes rather than evaluating all possible sizes. By selecting a limited set of candidate sizes (e.g., common GOP sizes like 16, 32, 64) and evaluating only those, the system performs partial action that achieves good rate distortion performance without the excessive computational cost of evaluating every possible GOP size, thus balancing performance improvement with acceptable device complexity.