Prediction-based image processing
US8311353B2 · kind B2 · utility
Assignee
Inventors
Key dates
| Filing date | May 6, 2008 |
| Grant date | Nov 13, 2012 |
| Priority date | — |
| Expiry date | Nov 17, 2028 |
Classification
- Technology area (CPC G)Physics
- CPC primaryG06T9/00
- WIPO fieldComputer technology
- WIPO sectorElectrical engineering
Abstract
A pixel block (300) is compressed by selecting a start depth value and a restart depth value based on the multiple depth values of the pixels (310-318) in the block (300). A respective plane representation (430) indicative of which plane of a start or restart depth value plane is determined for the pixels (311-318). These representations (430) are employed for selecting a pixel set comprising at least one other pixel (312, 314-317) in the block (300) for a pixel (318) to be encoded. The depth value(s) of the pixel(s) (312, 314-317) in the set are used for determining a prediction of the depth value of the pixel (318). The depth value and the prediction are employed for calculating a prediction error, which is encoded. The compressed pixel block (400) comprises the encoded prediction errors (460), a start value representation (420), a restart value representation (430) and the plane representations (440).
Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.