New class of feature oriented motion estimation for motion pictures

Problems that arise from using fast and full search algorithms cause poor motion compensated prediction along with some desirable feature, to which the human visual system is sensitive. A generalized class of algorithms to resolve the problems are proposed. The algorithms are adaptive, which include to consider the characteristics of block motions for typical image sequences, and to have an intelligent classifier to separate blocks containing different features. The motion vectors of these blocks are computed using feature frames and a masking factor. Results showed that this approach gives an improvement in accuracy for motion compensated frames and computational complexity, in comparison with intensity-based block motion estimation methods.