Method of automatically identifying and modifying the appearance of an object in successive frames of a video sequence
US6704045B1 · kind B1 · utility
Assignee
Inventor
Key dates
| Filing date | Sep 12, 1997 |
| Grant date | Mar 9, 2004 |
| Priority date | — |
| Expiry date | Sep 12, 2017 |
Classification
- Technology area (CPC H)Electricity
- CPC primaryH04N5/52
- WIPO fieldAudio-visual technology
- WIPO sectorElectrical engineering
Abstract
The invention provides a method of digitally processing a sequence of video frames, in which an object is selected in a first frame; digital object data corresponding to the shape of the object is stored on a digital data storage mechanism, together with digital modification data corresponding to a desired modification of at least one appearance attribute of the object; the object is identified in a second frame of the sequence by means of the object data; and the appearance attribute of the object in the second frame is modified in accordance with the stored modification data. The invention also provides a method of analysing a sequence of video frames, in which a plurality of corresponding lines are sampled from a plurality of consecutive frames of the sequence; the content of the sampled lines is compared for corresponding lines in different frames; a measure of the difference in the content of the corresponding lines is calculated; a first signal is produced if the measure of difference exceeds a first predetermined threshold; and a second signal is produced, in the absence of the first signal, if the measure of difference exceeds a second predetermined threshold, lower than th…
Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.