Fast motion estimation using UESA, threshold-half-stop and adaptive partial sum scan from gradient magnitude

In the fields of low bit rate coding and real time video coding, the three-step search (TSS) algorithm for fast motion estimation has played a key role because of its regularity, reasonable performance, and reduced amount of computation for searching the motion vector. In this work, a new fast TSS algorithm is proposed, which is based on UESA, half-stop method in matching pursuit and gradient-assisted adaptive matching scan for fast motion estimation. The first approach efficiently reduces checking points in the search range with reasonable PSNR performance. Two other algorithms remove unnecessary computations in obtaining the matching error without any degradation of error performance. This work is useful in the applications of software-based real-time video coding and low bit rate video coding.