A locally adaptive subsampling algorithm for software based motion estination

Motion estimation is an important technology in video coding system. Although many algorithms have been proposed to reduce its computational complexity, they are still insufficient for implementation of a software based system. This paper proposes several techniques for low-complexity motion estimation, such as a locally adaptive subsampling. Since they reduce the number of SAD operations, the computational complexity can be dramatically reduced while maintaining high image quality. Simulation results show that the proposed algorithm has about 20 times the speedup than TSS. This means that the proposed algorithms can attain enough processing performance for real-time implementation on embedded processors. Furthermore, PSNR of our method is about 0.2dB superior to that of TSS.