Palo Alto Research Center Inc.

Xerox Corporation

Jindong Chen

Lichan Hong

Eric Saund

David L Hecht

Stuart Kent Card

Ammar T Degani

Yizhou Wang

Jeff Breidenbach

Susan Evans

Ruth E Rosenholtz

John D Halbur

W Dale Knutsen

A method employing a hybrid classification model is used to perform optical character recognition operations for an image. Image data from the image is provided to a generative classification model of the hybrid model, and generative image classifications operations are performed, generating a feature data set which is outputted from the generative classification model. This feature data set is then provided to the discriminative classification model, and discriminative classification operations are performed to generate a classification of the image.

Innovative OCR systems and methods that combine a template based generative model with a discriminative model

A document processing system and method thereof are disclosed. The system comprises machine-readable data such as DataGlyphs and human-readable data such as rendering text of the DataGlyphs that are applied on the document. Processing devices of the system can decode document-processing instruction(s) embedded in the machine-readable data, and then optionally execute at least part of the instruction automatically. A designed stamp is also provided to generate the machine-readable data and human-readable data on the document.

Document processing devices, systems and methods thereof

Page turning of a three-dimensional electronic book is modeled as a dynamic animation. The page to be turned may be rotated by rotating it about a base line, and deformed by curling points of the front face of the turning page around the outer side of an imaginary geometric shape.

Systems and methods for turning pages in a three-dimensional electronic document

A system and method for improving the appearance of a raster image that includes glyph shapes is provided. A pair of glyph shapes is defined. Each glyph shape is differentiated two-dimensional machine readable data of relatively uniform appearance. Each of the glyph shapes is assigned to a different bit value. Digital data from an input source is encoded into a bitmap image space by representing a bit value for each item of the digital data as the glyph shape corresponding to that bit value. A raster image of the encoded glyph shapes is generated from the bitmap image space.

System and method for improving an appearance of a raster image comprising glyph shapes

A method for reducing the obtrusiveness of artifacts appearing in a glyph shape code image space includes encoding digital data values as a glyph shape code, such that the shape code is composed of glyphs having shapes that encode digital data values. Each distinct data value that is encoded is represented by the shape of a glyph, and these glyphs are spatially distributed in a bitmap image space. The bitmap image space is analyzed to determine the presence of artifacts within the bitmap image space, and a raster image is generated in which at least a single dot is placed within at least one of the artifacts in the image space.

Method for maintaining uniform spatial frequency over a dataglyph block

A method for reconstructing three dimensional shapes from two dimensional image domains uses glyph address carpets as a structured light pattern. The glyph address carpet includes a glyph encoding scheme in which the coordinates of any given point can be computed by looking at the glyph marks in a small neighborhood. The method includes calibrating at least one projecting device and at least one image detecting device and projecting a glyph address carpet pattern onto a three dimensional figure. The projected glyph address carpet pattern is detected and used to reconstruct the three dimensional coordinates for each readable glyph mark within the projected glyph address carpet pattern.

Three-dimensional active vision with glyph address carpet

In a system where the color space conversion process uses color profile look-up tables to select a profile for each of the possible attribute value combinations, assigning a value of wildcard to one or more attributes. In this case, if all of the values of one of the attributes can not be exactly matched, and that attribute has a wildcard value, the wildcard value is selected. The profile associated with the wildcard is designed to enable a reasonable color conversion in any case where an attribute calls for a particular value and the printer does not have a corresponding profile. The result is that the printer will most likely still print a reasonable image in the case where there is no matching profile.

Simple mechanism for profile selection

A method and system to print selective pages of color images within a print job having a plurality of pages using selected color improvement options such as black overprint, anti-aliasing, and trapping. The system identifies and selects one or more pages within a single print job having a set of a plurality of pages and is able to apply a plurality of different color improvement operations upon printing pages having color images thereon. A user interface enables the assigning of a first color improvement operation to a first subset of pages within the single print job. A processor determines if the page being processed is a member of the first subset of pages that have the first color improvement assigned thereto and performs the first color improvement operation upon the image data associated with the page when it determines that the page being processed is a member of the first subset of pages that have the first color improvement assigned thereto.

Growing community of inventors

Belmont, CA, United States of America

Jindong Chen