Temporal compensated motion estimation with simple block-based prediction

Motion compensated video format conversion (MC-VFC) needs true motion vectors, which have such features as spatial consistency and temporal extension. Based on these features, a temporally compensated motion estimation algorithm with simple block-based prediction (TC-SBP) is presented. Making use of the spatial-temporal correlation, the simple block-based prediction needs only three neighboring candidate vectors for each block, so that its computational complexity is highly reduced. At the same time, three temporally compensated updating candidates are used to accelerate the convergence of the algorithm efficiently for the global motion field. Measured using criteria relevant to MC-VFC and video processing applications, the new TC-SBP algorithm is shown to have a superior performance over alternative ones.