An efficient block matching algorithm based on a valid assumption of the convex distortion

This paper describes a new block matching algorithm (BMA), called optimal joint coordinate (OJC) search method, for motion-compensated coding. This method is in conjunction with seeking an optimal match between the pixels of a block in the current frame and the pixels of a corresponding block in the previous frame. Under the mean absolute error (MAE) distortion measure, this method completes searches in logarithm steps. In each step, OJC obtains two local optimum points within a corresponding search area. Also, the method determines a shrinked search quadrant, in which two local optimum points reside, for the next step. Simulation results demonstrate that our OJC search method is an optimal motion estimator. Its overall performance reaches a point where excellent convergence efficiency, high motion accuracy, and low information overhead meet