Global Search Using Partial Sampling and Alternate Pattern Blocks

摘要

The present invention relates to a partial sampling method and a global search method using alternating pattern blocks that effectively reduce the enormous amount of computation required for predicting block motion in integer pixel units and provide satisfactory performance. The first step of assigning a predetermined label to the pixels of the candidate search area CRS and the current image (hereinafter referred to as CP), and after performing the first step, selects a predetermined candidate position from the CRS. A second step of determining and storing the similarity with the pattern block which is a collection of pixels having the same label with respect to the candidate position; and after performing the second step, calculating the similarity with the pattern block at the current candidate position for all candidate motion vectors. And a third step of storing candidate positions of good similarity and outputting a motion vector. Similar to the method, but the candidate motion vector is calculated using a relatively small number of pixels, the amount of computation is significantly reduced.