A Fast Motion Estimation Algorithm Based on The Direction of Motion Vectors

This paper suggests a simple scheme in which the predicted motion vector is calculated using the spatio-temporal neighborhood information. The predicted motion vector is chosen as initial search center. The search pattern can be divided into 4 or 8 sectors based on the direction of the predicted motion vector. This predictive search center is found closer to the global minimum. The final sector based search pattern is adaptive as it is direction based and hence significantly reduces the computational complexity. Experimental results show that the proposed algorithm has good predicted image quality and less computational complexity as compared to other approaches.