This paper presents a new approach of block-based motion vector estimation, called Background/Foreground motion search (B/FS) algorithm. This algorithm separates the bbackground and foreground content inside a block to determine the motion vector of the block. As a result, the motion-compensated frames can be subjectively close to the original image. Experimental result shows that our approach provides a significant improvement in the subjective quality of motion-compensated frames as compared with traditional MSE/MAD-oriented block-based methods.
展开▼
