Method and apparatus for improving image appearance
AssigneeXerox Corporation
InventorsMarshall W. Bern, David Goldberg
Life of this patent
Estimating the term…
Abstract
Connected components of dark pixels are clustered from across the image. A “most likely” representative image for each cluster of images is determined, with likelihood determined by a probabilistic model of the image capturing process. An a priori (prior) probability distributions on bitmaps may be used to determine the most likely representative images. For example, a priori probability distributions based on so-called chain codes are implemented. The representative images are used to cluster connected components. Clustering may be repeated. The output page is assembled by replacing each member of a cluster of images by that cluster's representative image.
Claims30
A method for improving appearance of captured bilevel image data, comprising:
- receiving a degraded input bilevel image;
- connecting dark pixels in the input image to adjacent dark pixels to form connected components comprising a set of dark pixels connected either diagonally or orthogonally and surrounded by white pixels;
- performing initial clustering of individual connected components;
- determining a “most likely” cluster representative by use of a probabilistic model of the scanner used for scanning;
- and
- assembling the sets by substituting the “most-likely” cluster representative for each family member of each cluster set to form an output image,
- wherein the step of performing initial clustering includes pair-wise matching connected components and determining a match if the pair are within a certain threshold of matching to form cluster sets, each with one or more family members formed of individual connected components,
- the step of determining a “most likely” cluster representative for each cluster set uses a optimization procedure in which at least one iteration of pixel flipping is performed,
- multiple iterations of pixel flipping are performed, with a first iteration of pixel flipping being performed with a first threshold, and
- a subsequent iteration being performed with a more progressive threshold being applied until a maximized probability is attained.
A method for improving appearance of captured bilevel image data, comprising:
- receiving a degraded input bilevel image;
- connecting dark pixels in the input image to adjacent dark pixels to form connected components comprising a set of dark pixels connected either diagonally or orthogonally and surrounded by white pixels;
- performing initial clustering of individual connected components;
- determining a “most likely” cluster representative by use of a probabilistic model of the scanner used for scanning;
- assembling the sets by substituting the “most-likely” cluster representative for each family member of each cluster set to form an output image;
- and
- reclustering by comparing cluster representatives of clusters, wherein
- reclustering stops when the larger cluster sets have fewer than four family members.
Dependencies are inferred from the claim text.