Motion estimation by a new and adaptive hierarchical phase correlation

摘要

The motion data in a video is a very important item to save or transmit the video. It decreases the storage area and transmission bandwidth. So the motion estimation issue becomes very important in the video technology. The present invention about motion estimation is based on the frequency domain and it focuses on the phase correlation of the frame sequences. The main idea behind this invention is to compare the ratio between the first highest peak and second highest peak with an adaptive threshold value which changes itself according to the result of the phase correlation. If the ratio is lower, there are two or more motions included in the block and it should be further split. Then the splitting procedure continues until the size of blocks becomes equal to appropriate size. The motion vector corresponding to the highest peak is assigned to the entire frame. The new method provides a better PSNR performance with less motion vectors and automatically adapts itself to the frames.