RATE-DISTORTION OPTIMAL MOTION ESTIMATION ALGORITHMS FOR MOTION-COMPENSATED TRANSFORM VIDEO CODING

Motion estimation and compensation is widely used for exploiting tempo ral correlation within an image sequence. To find motion vectors that lead to high compression, most motion estimation approaches use a sour ce distortion measure, such as mean-square error (MSE) or mean-absolut e error (MAE), as a search criterion, When incorporated into a closed- loop motion-compensated (MC) transform video coder, these schemes prod uce noisy motion fields which significantly increase the bit-rates req uired to represent motion vectors, In view of this problem, this paper presents a rate-distortion optimal motion estimation algorithm, The p roposed scheme improves rate performance of the estimated motion field while maintaining the peak signal-to-noise ratio (PSNR) prediction qu ality of the distortion-based methods, thereby enabling an efficient b it allocation between motion information and transform-coded predictio n residuals, For coders in which motion vectors are differentially enc oded, the rate-distortion optimization process is formulated as a shor test-path-finding problem. Adopting this framework, we show that the o ptimal solution for the conventional block-based motion estimation, fo llowed by one-dimensional (1-D) differential coding and Huffman coding , can be obtained by using dynamic programming or the Viterbi algorith m, Since memory and computational requirements for the optimal algorit hm increase dramatically with the search range, we propose an effectiv e fast algorithm that closely approximates the optimal performance whi le requiring considerably less complexity, When used in a hybrid MC co ding system, the proposed scheme outperforms the conventional MSE- or MAE-optimal motion estimation approach in terms of PSNR performance un der a low-bit-rate constraint, Our experimental results demonstrate ov erall gains in the range of 0.3-1.5 dB.