Fast Motion Estimation Algorithm for Scalable Motion Coding

Motion information scalability is an important requirement for a fully scalable video codec, a novel and fully scalable motion model (SMM) is proposed to enable a scalable video codec to achieve optimality over a wide range of bit rate, resolution, and frame rate. In SMM, Full search algorithm for motion estimation can find the best matching region in the search area of the reference frame. However, the computational complexity is very high. To reduce complexity, this paper proposes a fast motion estimation algorithm based on an improved particle swarm optimization (PSO) for scalable motion coding, which uses early termination of search and improves initial population and partial particle mutation. Experimental results show that the proposed algorithm provides faster speed with fewer search points while the average PSNR is almost close to full search algorithm with little loss.