Abbyy Development Inc.

Abbyy Software Limited

Abbyy Infopoisk LLC

Diar Tuganbaev

Konstantin Zuev

Irina Zosimovna Filimonova

Konstantin Anisimovich

Vladimir Pavlovich Selegey

Sergey Zlobin

Maryana Skuratovskaya

Dmitri Deriaguine

A method for processing a batch of scanned images is provided. The method comprises processing the scanned images into documents. For documents of multiple pages, the method comprises maintaining a page-based coordinate system to specify a location of structures within a page and joining the pages to form a multi-page sheet having a sheet-based coordinate system to specify a location of structures within the multi-page sheet. The method comprises performing a data extraction operation to extract data from each document, said data extraction operation including a page mode wherein structures are detected on individual pages using the page-based coordinate system and a document mode wherein structures are detected within the entire document using the sheet-based coordinate system.

Flexible structure descriptions for multi-page documents

In one embodiment, the invention provides a method for machine translation of a source document in an input language to a target document in an output language, comprising generating translation options corresponding to at least portions of each sentence in the input language; and selecting a translation option for the sentence based on statistics associated with the translation options.

Deep model statistics method for machine translation

The invention involves a method for processing of machine-readable forms or documents of non-fixed format. The method makes use of, for example, a structural description of characteristics of document elements, a description of a logical structure of the document, and methods of searching for document elements by using the structural description. A structural description of the spatial and parametric characteristics of document elements and the logical connections between elements may include a hierarchical logical structure of the elements, specification of an algorithm of determining the search constraints, specification of characteristics of every searched element, and specification of a set of parameters for a compound element identified on the basis of the aggregate of its components. The method of describing the logical structure of a document and methods of searching for elements of a document may be based on the use of the structural description.

Method for object recognition and describing structure of graphical objects

Methods for processing machine-readable forms or documents of non-fixed format are disclosed. The methods make use of, for example, a structural description of characteristics of document elements, a description of a logical structure of the document, and methods of searching for document elements by using the structural description. A structural description of the spatial and parametric characteristics of document elements and the logical connections between elements may include a hierarchical logical structure of the elements, specification of an algorithm of determining the search constraints, specification of characteristics of searched elements, and specification of a set of parameters for a compound element identified on the basis of the aggregate of its components. The method of describing the logical structure of a document and methods of searching for elements of a document may be based on the use of the structural description.

Object recognition and describing structure of graphical objects

In one embodiment, the invention provides a method, comprising detecting data fields on a scanned document image; generating a flexible document description based on the detected data fields, including creating a set of search elements for each data field, each search element having associated search criteria; and training or modifying the flexible document description using, for example, a search algorithm to detect the data fields on additional training images based on the set of search elements.

Creating flexible structure descriptions

Embodiments of the invention disclose techniques for processing of machine-readable forms of unfixed or flexible format. An auxiliary brief description may be optionally specified to determine the spatial orientation of the image. A method of searching for elements of a document comprises the following main operations in addition to the operations of preliminary image processing: selecting the varieties of structural description from several available variants, determining the orientation of the image, selecting the text objects, where the text must be recognized, and determining the minimal required volume of recognition, recognizing the text objects, searching for elements of the form. Searching for elements of the form comprises the following actions: selecting a searched element in the structural description, gaining the algorithm of search constraints from the structural description, searching for the element, testing the obtained variants.

Methods of object search and recognition

In one embodiment, there is provided a method for an Optical Character Recognition (OCR) system. The method comprises: recognizing an input character based on a plurality of classifiers, wherein each classifier generates an output by comparing the input character with a plurality of trained patterns; grouping the plurality of classifiers based on a classifier grouping criterion; and combining the output of each of the plurality of classifiers based on the grouping.

Accuracy of recognition by means of a combination of classifiers

Techniques and methods are disclosed herein for combining and weighting of values from and associated with classifiers. Classifiers are used to recognize characters as part of an optical character recognition (OCR) system. Various methods of normalization facilitate combining of results of classifiers. For example, weight values may be entered into a weight table having two columns, one that includes weights from comparing patterns with images of correct characters, the other column includes weights from comparing patterns with images of incorrect characters.

Classifier combination for optical character recognition systems utilizing normalized weights and samples of characters

A method for processing a batch of scanned images is provided. The method comprises processing the scanned images into documents; for documents comprising multiple pages maintaining a page-based coordinate system to specify a location of structures within a page and joining the pages to form a multi-page sheet having a sheet-based coordinate system to specify a location of structures within the multi-page sheet; performing a data extraction operation to extract data from each document, said data extraction operation comprising a page mode wherein structures are detected on individual pages using the page-based coordinate system and a document mode wherein structures are detected within the entire document using the sheet-based coordinate system.

Data capture from multi-page documents

A method for processing a batch of scanned images is disclosed. The method includes processing the scanned images into documents. For documents of multiple pages, the method maintains a page-based coordinate system to specify a location of structures within a page and joins the pages to form a multi-page sheet associated with a sheet-based coordinate system to specify a location of structures within the multi-page sheet. Data may be extracted from each document through a page mode wherein structures are detected on individual pages using the page-based coordinate system and a document mode wherein structures are detected within the entire document using the sheet-based coordinate system.

A method of identifying stricken-out characters in handwriting, comprising parsing a scanned image into regions and objects, defining objects containing handwritten characters, applying structural or feature classifiers for primary character recognition, applying one or more supplemental feature classifiers preliminarily trained by strike-out characters, and identifying a stricken-out character if any. The stricken-out character may be further examined by special procedures, either automated or manual.

Growing community of inventors

Moscow, Russia

Diar Tuganbaev