Cost-effective multiresolution motion estimation algorithm for rate distortion optimized high definition video encoder

This paper proposes an efficient multiresolution motion estimation algorithm (MMEA) well-suited for high definition video encoder VLSI implementation. 16-way parallel process element arrays achieve fast search with regular data access and efficient data reuse. Moreover, the parallel arrays can be efficiently reused at three hierarchical levels for sequential motion vector (MV) refinement. Also, an intelligent algorithm for multiple candidate MV selection is proposed to compensate for performance degradation due to parallel MMEA. The whole algorithm achieves a good balance between complexity and performance. Simulation results show that only 0.14dB PSNR degradation at the worst case is achieved compared with full search algorithm.