Study Of Motion Estimation Algorithm In Video Compress

With the fast develop of multimedia techniques, there are many areas need more efficient real-time compression techniques for digital image processing. Fast compression algorithm is the key for this problem.Motion estimation is rather complex in computation and spends much time in the inter-frame video compression. Fast motion estimation has always been a favorite to the researchers in the field. In this paper, we first analyse some popular motion estimation algorithms, some of them determine ,the search direction too early, and easily fall into the local minimum point, so precision is affected. The motion vectors of the blocks within this group must have some relationship between them. If the relationship between the motion vectors of the group of blocks can be identified, it is then possible to predict the target motion vector from the motion vectors of the group of blocks. So we could use the predicted motion vector as ours starting point. Based this method of predicting starting point, we propose a fast motion estimation algorithm based on center search to decrease the computation of trivial motion in DS. Searching pattern adapts to trivial motion and the method of predicting starting point adapts to furious motion are designed to satisfy requirements on both speed and accuracy. The simulation results show that the searching speed and accuracy of the algorithm are superior to the recently developed excellent algorithm DS. Because the precision and computation of motion estimation are relevant to block matching criterion, so we present a new criterion called SC-SAD which enhances the conventional SAD function by adding a smoothness control for the residue block with adding few computation. Simulation results show that by applying the new criterion in motion estimation, both the improvement in precision and reduction in bit rate can be achieved. In the process of motion estimation, when we get the final matched block, we can predetermine the blocks whose quantized DCT coefficients are completely zero. It is applied to eliminate redundant computations of DCT transform and quantization. A new method of predetermining the blocks whose quantized DCT coefficients are completely zero is proposed in the paper. Experimental results show that the improved method can determine the blocks whose quantized DCT coefficients are completely zero effectively, and it is very useful for reducing the complexity and computation in video coder.