Back to results
US 12,694,703 B2Granted
Method and apparatus employing context-aware optical character recognition to extract form information
AssigneeKonica Minolta Business Solutions U.S.A., Inc.
InventorTing Xu
Life of this patent
Estimating the term…
Abstract
Method and apparatus for context-aware character recognition, using surrounding materials from ambiguous characters to identify the ambiguous characters correctly. There is better optical character recognition (OCR) accuracy for intrinsically confusing characters, printed or handwritten, and increased information extraction accuracy for scanned forms, including forms filled out by hand.
Claims20
A method comprising:
- receiving a text-containing document into a deep learning system;
- using the deep learning system, detecting lines in the text-containing document;
- using the deep learning system, responsive to the detecting, segmenting text in the text-containing document into one or more regions;
- using the deep learning system, identifying one or more keywords in each of the one or more regions;
- using the deep learning system, responsive to the identifying of the one or more keywords, identifying content for each of the one or more keywords;
- using the deep learning system, constructing a new image with the one or more keywords and the corresponding content;
- and
- using the deep learning system, responsive to the constructing, outputting form fields as serialized text.
5 more elements
An apparatus comprising:
- a deep learning system comprising at least one processor and a non-transitory memory that contains instructions that, when executed, enable the deep learning system to perform a method comprising:
- responsive to receipt of a text-containing document, detecting lines in the text-containing document;
- responsive to the detecting, segmenting text in the text-containing document into one or more regions;
- responsive to the segmenting, identifying one or more keywords in each of the one or more regions;
- responsive to the identifying of the one or more keywords, identifying content for each of the one or more keywords;
- constructing a new image with the one or more keywords and the corresponding content;
- and
- responsive to the constructing, outputting form fields as serialized text.
5 more elements
Dependencies are inferred from the claim text.