Motion Estimation Method
Motion Estimation Method
Block motion estimation algorithms (BMA) are widely adopted by video coding standards such as MPEG-1, MPEG-2, MPEG-4, H.261, H.263 and the emerging H.264/AVC, mainly due to their simplicity and good distortion performance. The most straightforward BMA is referred to as full search (FS), which exhaustively evaluates all possible candidate blocks within the search window. It has been estimated that the computation of FS could consume up to 70% of the total computation of the video encoding process. To solve this problem, many fast algorithms have been developed to speed up the motion estimation process. However, search point pattern based fast BMAs are easily trapped in local minima.
To tackle the above problem, a novel directional gradient descent search (DGDS) is developed. In addition, a Fast DGDS (FDGDS) algorithm is also developed to further improve the speed of DGDS. Experimental results show that the technologies mentioned above can lead to a significant improvement in computation reduction, as compared with well-known fast block motion estimation algorithms. In addition, FDGDS algorithms can achieve a higher speed, compared with UMHexagonS algorithm in H.264/AVC implementation, and at the same time maintain very similar coding performance.
All digital video encoding devices such as video recorder, digital camera and video camera, and high-definition video system, can use the FDGDS algorithms to reduce cost and power consumption The market is huge in both portable devices and home entertainment products
14/06/2008
12/139,469
香港
