EFFICIENT COST MEASURES FOR MOTION ESTIMATION AT LOW BIT RATES

We present and compare methods for choosing motion vectors for block-b ased motion-compensated video coding. The primary focus is on videopho ne and videoconferencing applications, where low bit rates are necessa ry, where motion is usually limited, and where the amount of computati on is also limited. In a typical block-based motion-compensated video coding system, motion vectors are transmitted along with a lossy encod ing of the residuals. As the bit rate decreases, the proportion requir ed to transmit the motion vectors increases. We provide experimental e vidence that choosing motion vectors explicitly to minimize rate (incl uding motion vector coding), subject to implicit constraints on distor tion, yields better rate-distortion tradeoffs than minimizing some mea sure of prediction error. Minimizing a combination of rate and distort ion yields further improvements. Although these explicit-minimization schemes are computationally intensive, they provide invaluable insight which we use to develop practical algorithms. We show that minimizing a simple heuristic function of the prediction error and the motion ve ctor code length results in rate-distortion performance comparable to explicit-minimization schemes while being computationally feasible. Ex perimental results are provided for coders that operate within the H.2 61 standard.