Gain correction for encoding video images
AssigneeIntel Corporation
InventorsThomas R. Gardos, Joe Monaco
Life of this patent
Estimating the term…
Abstract
A background region of the current image is compared to a reference region to characterize the gain associated with the current image. The values of the unsaturated background pixels of the current image are then adjusted to correct for the characterized gain before encoding the current image. In a preferred embodiment, the reference region is generated based on the previous n frames in the video sequence.
Claims18
A computer-implemented process for processing video signals, comprising the steps of:
- (a) selecting a background region of a current image of a video stream;
- (b) generating a reference region using regions corresponding to background regions of one or more images of the video stream;
- (c) characterizing gain of the current image based on the reference region and the background region;
- and
- (d) generating a gain-corrected image by correcting the current image based on the gain.
An apparatus for processing video signals, comprising:
- (a) means for selecting a background region of a current image of a video stream;
- (b) means for generating a reference region using regions corresponding to background regions of one or more images of the video stream;
- (c) means for characterizing gain of the current image based on the reference region and the background region;
- and
- (d) means for generating a gain-corrected image by correcting the current image based on the gain.
A storage medium encoded with machine-readable computer program code for processing video signals, wherein, when executed by a computer, the computer program code causes the computer to:
- (a) select a background region of a current image of a video stream;
- (b) generate a reference region using regions corresponding to background regions of one or more images of the video stream;
- (c) characterize gain of the current image based on the reference region and the background region;
- and
- (d) generate a gain-corrected image by correcting the current image based on the gain.
Dependencies are inferred from the claim text.