A simplified motion estimator based on binary correlation

摘要

Motion estimation plays an important part particularly for source encoding with motion compensating prediction of moving pictures. Due to the large amount of hardware, application is limited to systems requiring high coding efficiency. In this paper a method is proposed which can reduce complexity of the motion estimator by about a factor six compared to conventional solutions. The crucial point is that the two input signals of the motion estimator are roughly quantized before searching for the optimum displacement vector resulting in a very simple arithmetic-logic-unit and smaller memories. The method is analyzed mathematically and its good performance is demonstrated by computer simulations at bit rates from 64 to about 300 kbit/s using moving pictures.