Konica Minolta Laboratory U.s.a., Inc.

Konica Minolta Business Solutions U.s.a., Inc.

Darrell Eugene Bellert

Eduard G Bauboeck

Tim Prebble

A method, non-transitory computer readable medium, and system for inferring certain texts as stylized section headings in an electronic document (ED). Stylized section headings are section headings that have unique styling distinct from the body of text below each stylized heading. In particular, the stylized section headings are identified based on styling information in the ED. Identifying stylized section headings includes grouping candidate headings based on identification of dominant styling, locating high level fragments, and repeatedly locating nested fragments from within higher level fragments. The ED may or may not include explicitly identified headings in the document.

Identifying section headings in a document

A method for processing an electronic document (ED) to infer a sequence of section headings in the ED. The method includes generating, by a computer processor, based on regular expression matching of a predetermined section heading pattern and a plurality of characters in the ED, a list of candidate headings in the ED; generating, by the computer processor and based on the list of candidate headings, a list of chain fragments for inferring a portion of the sequence of section headings; and generating, by the computer processor and based on predetermined criteria, the sequence of section headings by merging at least two chain fragments in the list of chain fragments.

Identifying sequence headings in a document

An image processing method to generate a layout of searchable content from a physical document. The method includes generating extracted content blocks in the physical document, generating, based on a bounding box of a text block, a layout rectangle that identifies where machine-encoded text is placed in the layout of the searchable content, generating, based on a bounding box of a non-text block, an avoidance region that identifies where the machine-encoded text is prohibited in the layout of the searchable content, generating, based on the layout rectangle and the avoidance region, a draft layout of the searchable content, and iteratively adjusting a point size of the machine-encoded text in the draft layout to generate the layout of the searchable content.

Approximating the layout of a paper document

A method for processing an electronic document (ED) to infer columns in the ED, where the ED comprises a plurality of characters. The method includes generating a mark-up version of the ED having text-layout attributes of the characters in the ED, where the characters are grouped into paragraphs based on the text-layout attributes, and each paragraph corresponds to a paragraph bounding box surrounding a corresponding paragraph, generating border pieces by initiating a pair of left scan and right scan from each paragraph bounding box to identify any adjacent paragraph bounding box, and generating, based at least on the border pieces, column borders for use in inferring the columns in the ED, where at least one column has a vertically aligned portion of the paragraphs.

Column inferencer based on generated border pieces and column borders

A method for processing an image including a table is provided. The method includes: identifying a plurality of connected components that form the table; generating a skeleton graph comprising a plurality of edges based on the plurality of connected components; calculating a typesetness score that compares the table to a template table based on the plurality of edges; determining a rotation angle for the table and generating a custom coordinate system for the table based on the rotation angle; identifying a plurality of cells in the table based on the skeleton graph, the typesetness score, and the custom coordinate system; linking each of the plurality of cells based on the plurality of edges and validating a link between the plurality of cells; generating a grid of cells based on the link and comprising the plurality of cells; and exporting the grid as a high-level representation of the table.

Identifying hand drawn tables

A method for processing an electronic document (ED) to infer titles in the ED is provided. The method includes: generating a mark-up version of the ED comprising text-styling attributes, text-layout attributes, and text content information of characters included in the ED; generating statistical information of the text-styling and text-layout attributes; calculating, for each text-styling and text-layout attribute, a relative weight score; calculating, for each paragraph in the ED: a styling criteria score and a layout criteria score based on the statistical information and the relative weight scores; a text content score based on the text content information; and a title confidence score based on the styling criteria score, the layout criteria score, and the text content score; and generating a metadata for the ED that includes the title confidence score for each paragraph for use in inferring the titles in the ED.

Title inferencer

A method for image processing is provided. The method includes: obtaining an image including a table; generating, for the table, a skeleton graph including a plurality of edges; identifying a plurality of angles and a plurality of lengths for the plurality of edges; and calculating a typesetness score that compares the table to a template table based on the plurality of angles and the plurality of lengths.

Typesetness score for a table

A method for processing an image that includes a table is provided. The method includes: obtaining a set of edges associated with a cell candidate of the table; determining a longest edge in the set of edges and a first orientation of the longest edge; determining, in the set of edges, an initial edge for a first cell border comprising the first orientation and the longest edge; building the first cell border based on the initial edge, the first orientation, and the set of edges; calculating a second orientation for a second cell border by rotating the first orientation; building the second cell border based on an initial edge for the second cell border and the second orientation; and validating the cell candidate as a cell of the table in response to building the first cell border and building the second cell border.

Table cell validation

A method for image processing is provided. The method includes: obtaining an image including a table that includes a plurality of cells that each includes a first border, a second border opposite the first border, a third border, and a fourth border opposite the third border; identifying, on a skeleton graph generated for the table, a first non-corner vertex on the first border of a first cell of the plurality of cells; determining a first plurality of splitlines extending from the first non-corner vertex and intersecting with the second border of the first cell without intersecting with the third or fourth border of the first cell; calculating a first plurality of confidence values for the first plurality of splitlines; and splitting the first cell by generating a first artificial edge, based on the first plurality of confidence values and a splitline of the first plurality of splitlines.

Splitting merged table cells

A method for image processing is provided. The method includes: generating a skeleton graph associated with the table and comprising a plurality of edges; identifying, in the skeleton graph, a corner vertex, a starting vertex adjacent to the corner vertex, and an ending vertex adjacent to the corner vertex; determining a set of route options for the starting vertex that includes a first set of vertices adjacent to the starting vertex; selecting a candidate vertex from the first set of vertices as a first vertex; determining a set of route options for a second vertex comprising a second set of vertices adjacent to the second vertex; determining the second set of vertices comprises the ending vertex; and generating a route for a cell in the table and comprising the corner vertex, the starting vertex, the first vertex, the second vertex, and the ending vertex.

Detection of near rectangular cells

A method for image processing is provided. The method includes: obtaining an image including a table; generating a skeleton graph associated with the table and including a plurality of edges; identifying a first terminal vertex on the plurality of edges; generating a first region of interest for the first terminal vertex; determining a first target within the first region of interest; and repairing the table by adding a first artificial edge between the first terminal vertex and the first target to the skeleton graph.

Repairing tables

A method for image processing includes obtaining a mask from an image; identifying, in the mask, a first stroke segment and a second stroke segment for possible merging; determining that the first stroke segment comprises a first plurality of cross edges that overlaps a second plurality of cross edges in the second stroke segment; identifying a plurality of cross edge pairs from the first plurality of cross edges and the second plurality of cross edges; determining a count of the plurality of cross edge pairs that satisfy a maximum width tolerance value of the first stroke segment; and merging the first stroke segment and the second stroke segment based on the count.

Merging overlapping broken lines of a table

A method for image processing includes obtaining a mask of an image; identifying, in the mask, a first stroke segment and a second stroke segment for possible merging, wherein the first stroke segment and the second stroke segment do not overlap; calculating a plurality of merger eligibility scores for the first stroke segment and the second stroke segment; calculating an aggregate merger eligibility score for the first stroke segment and the second stroke segment based on the plurality of merger eligibility scores; and merging the first stroke segment and the second stroke segment based on the aggregate merger eligibility score.

Merging non-overlapping broken lines of a table

A method for image processing includes obtaining a mask of a stroke from an image; determining a plurality of cross edges for the stroke based on the mask; generating a histogram comprising a plurality of widths of the cross edges and a plurality of frequencies of the plurality of widths from the cross edges; estimating a lower bound of a width range for the stroke based on a mode width of the plurality of widths, a first subset of the plurality of widths below the mode width, and a first plurality of weights assigned to the first subset of the plurality of widths; and estimating an upper bound of the width range for the stroke based on the mode width, a second subset of the plurality of widths above the mode width, and a second plurality of weights assigned to the second subset of the widths.

Method for determining width of lines in hand drawn table

A method for image processing includes obtaining a mask of a stroke from an image and identifying a plurality of cross edges for the stroke based on the mask and a reference line. The plurality of cross edges includes a group of adjacent cross edges that intersect the reference line. The method further includes (a) calculating a first vector based on positions of at least two of the cross edges in the group, (b) expanding the group, based on the first vector, to include cross edges adjacent to the group that do not intersect the reference line, (c) calculating a second vector based on positions of at least two of the cross edges in the expanded group, and (d) expanding the expanded group, based on the second vector, to include a second group of adjacent cross edges nearby the expanded group that do not intersect the reference line.

Growing community of inventors

Boulder, CO, United States of America

Darrell Eugene Bellert