Removal of image artifacts from sensor dust
AssigneeMicrosoft Corporation
InventorsStephen Lin, Baining Guo, Changyin Zhou
Life of this patent
Estimating the term…
Abstract
Removal of the effects of dust or other impurities on image data is described. In one example, a model of artifact formation from sensor dust is determined. From the model of artifact formation, contextual information in the image and a color consistency constraint may be applied on the dust to remove the dust artifacts. Artifacts may also be removed from multiple images from the same or different cameras or camera settings.
Claims22
A method of processing an image including an artifact, the method comprising:
- obtaining the image;
- detecting the artifact in the image;
- and
- processing the image based at least in part on an artifact formation model wherein the artifact is removed from the image;
A method of constructing an artifact formation model over an image region, the method comprising defining a function space of possible dust artifacts α determined at least in part by a projection geometry factor τ and a bounds of dust transparency β.
One or more computer-readable media having computer-executable instructions stored thereon, the computer-executable instructions embodying a method of recovering a dust-free image from an image, the method comprising:
- partitioning the image into a plurality of overlapping regions;
- fitting of a Gaussian function to each of the plurality of overlapping regions;
- evaluating the concavity of each of the fitted Gaussian functions wherein one or more of the overlapping regions become candidate regions based on the corresponding concavity;
- estimating a projection geometry factor;
- optimizing an energy function for the candidate regions, the energy function based at least in part on the projection geometry factor;
- selecting one or more of the candidate region based at least in part on the corresponding energy function;
- and
- recovering a dust-free region for each of the one or more selected candidate regions according to minimized values of each of the corresponding one or more selected candidate region's energy function.
Dependencies are inferred from the claim text.