Quassi Lossless Motion Estimation Algorithm Using Fast Elimination of Checking Points

We propose an algorithm that reduces unnecessary computations, while keeping almost same prediction quality as that of the full search algorithm. In the proposed algorithm, we can reduce unnecessary computations efficiently by calculating initial matching error point from first partial errors. Additionally, we use different search strategy compared with conventional spiral search pattern. By doing that, we can increase the probability that hits minimum error point as soon as possible.