A New Adaptive Motion Search with Elastic Pattern Based on Motion Vector Field
Xing Li · Chinese Journal of Computers · 2004
A new adaptive motion search algorithm with elastic pattern based on motion vector field is presented in this paper. Based on high spatial-temporal relativity of the video sequences, the algorithm started with the uniformity of motion vector field, combining a series of skills such as fast block matching,selective strategy of adaptive elastic pattern,adaptive threshold and effective half-stop criteria. It is an adaptive fast motion estimation algorithm with scalable structure. The experiment result shows that the algorithm could obtain almost the same PSNR as the full search algorithm at very low computation cost. Meanwhile, in terms of speed and quality, it apparently excels the newest fast motion estimation algorithm recommended by MPEG-4 Part 7. It has much better capability in low and medium bit-rate compression, thus, it is very suitable to be applied to the field of low-bit rate real-time video compression.