An Integrated Algorithm for Fractional Pixel Interpolation and Motion Estimation of H.264

Abstract

Fractional pixel motion compensation technology is an area in video image compression that can provide significant gains for coding efficacy, but this improvement comes the associated cost of high computational complexity. This additional complexity arises from two aspects: fractional pixel motion estimation (FPME) and fractional pixel interpolation (FPI). Different from current fast algorithms, we use the internal link between FPME and FPI as a factor in considering optimization by integrally manipulating them rather than attempting to speed them up separately. To coordinate with FPME and FPI, our proposed algorithm estimates fractional motion vectors and interpolates fractional pixels in the same order, which will satisfy the criteria of cost/performance efficiency. Compared with the FFPS+XFPI (the FPI method in X264), the proposed algorithm has already reduced the speed by a factor of 60% without coding loss. Furthermore, the proposed algorithm also achieves a much higher speed and better R-D performance than other fast algorithms e.g. CBFPS+XFPI. This integrated algorithm, therefore, improves the overall video coding speed by a significant measure and its idea of jointly optimizing the computational cost and the R-D performance can be extended to speeding up an even finer fractional motion compensation, such as 1/8 pixel, and to designing new interpolation filters for H.265.