Hybrid Inter Bi-Prediction with Bilateral Motion Vector Derivation

The patent applies self-service by enabling the decoder to autonomously derive one motion vector from candidate lists using bilateral matching criteria without requiring explicit signaling from the encoder. The decoder independently evaluates candidates based on matching errors between reference blocks, selecting the best motion vector candidate automatically, thus eliminating the need for transmitting this information and reducing signaling overhead while maintaining prediction accuracy.

The patent applies partial action by performing motion vector derivation only for one of the two motion vectors in bi-prediction, rather than both. The other motion vector is still explicitly signaled. This partial application of derivation reduces the overall computational burden compared to deriving both motion vectors, while still achieving signaling overhead reduction for the derived vector.

The patent applies segmentation by dividing the bi-prediction process into two distinct parts: one motion vector that is explicitly signaled and another that is derived using decoder-side techniques. This segmentation allows each part to be optimized independently - the signaled vector ensures accuracy where needed, while the derived vector reduces signaling overhead, together improving the rate-distortion tradeoff through differentiated handling of motion vector transmission.