Adaptive step-size motion estimation based on statistical sum of absolute differences
US6014181A · kind A · utility
Assignees
Inventor
Key dates
| Filing date | Oct 13, 1997 |
| Grant date | Jan 11, 2000 |
| Priority date | — |
| Expiry date | Oct 13, 2017 |
Classification
- Technology area (CPC H)Electricity
- CPC primaryH04N19/56
- WIPO fieldAudio-visual technology
- WIPO sectorElectrical engineering
Abstract
A novel motion estimation algorithm, AMESSAD (adaptive motion estimation based on statistical sum of absolute difference) is provided. The algorithm adaptively determines motion search step size based on statistical distribution of SAD (sum of absolute difference). That is, search step sizes to estimate motion in one portion of a frame are calculated using SAD values from neighboring portions of the frame. The efficient search procedure improves the implementation of motion compensation and transform based hybrid video coders, such as the H.26P and MPEG-X standard video compression. Compared with fixed step-size motion estimation, the adaptive algorithm improves motion estimation and hence overall video encoding speed. In addition, improved visual quality can be achieved in many cases because the algorithm differentiates regions with motion activity and allocates more motion estimation resources to local areas or local frames with higher motion content.
Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.