In predictive encoding of motion vectors, the fixed prediction scheme has been widely used, which adopts the motion vector of the last encoded block as the prediction vector. This paper proposes an adaptive prediction scheme which adaptively selects a prediction vector out of some candidate vectors. The selected vector requires less bits for encoding the resultant error vector than the others. The candidate vectors can be obtained from the motion vectors of the previously encoded blocks. We suggest a scheme obtaining the candidate vectors when the number of candidate vectors is set to 2. The overhead 1 bit is needed only when the two candidate vectors are different from each other. Simulation results show that the proposed prediction scheme reduces the number of bits for encoding motion vectors by about 10 percent when compared with the fixed prediction scheme.
展开▼