System for template based extracting information from an identity card
First Claim
Patent Images
1. A system for automatically extracting information from an identity document, the system comprising:
- a) at least one light source for illuminating the identity document;
b) at least one digital camera including a lens and a two-dimensional sensor array configured to acquire at least one two-dimensional image frame of the identity document; and
c) a processor for processing said at least one two-dimensional image frame, said processor being in communication flow with a document-template database, said processor is configured to apply an optical character recognition (OCR) routine for extracting textual information from said at least one two-dimensional image frame of the identity document in the form of alphanumerical information, wherein said OCR routine includes;
i) determining a type of the identity document;
ii) fetching a matching template from said document-template database, wherein said matching template corresponds to said determined type of the identity document, and wherein said template includes one or more data fields;
iii) overlaying said at least one two-dimensional image frame with said matching template, wherein each data field demarcates a respective data region of said overlaid image frame, to thereby obtain a geometrical correlation between said at least one two-dimensional image frame and said template;
iv) selecting data regions, wherein said textual information is extracted from selected data regions;
v) determining boundaries between text and background based on a minimum contrast between a character and other colored data on said image frame of the identity document, approximately within the boundaries of said selected data regions;
vi) extracting black and white glyphs using at least a portion of said determined boundaries; and
vii) applying said OCR routine on said black and white glyphs, thereby recognizing symbols and characters.
6 Assignments
0 Petitions
Accused Products
Abstract
The present invention provides a unique and novel system for acquiring digital image frames of identification documents such as driver'"'"'s license, passports, or medical insurance records using a digital camera so as to establish a high resolution image frame and extracting data automatically with machine vision tools so as to acquire accurate data. The present invention teaches also a system that acquires multi-spectral image frames of both sides of the identification document.
32 Citations
20 Claims
-
1. A system for automatically extracting information from an identity document, the system comprising:
-
a) at least one light source for illuminating the identity document; b) at least one digital camera including a lens and a two-dimensional sensor array configured to acquire at least one two-dimensional image frame of the identity document; and c) a processor for processing said at least one two-dimensional image frame, said processor being in communication flow with a document-template database, said processor is configured to apply an optical character recognition (OCR) routine for extracting textual information from said at least one two-dimensional image frame of the identity document in the form of alphanumerical information, wherein said OCR routine includes; i) determining a type of the identity document; ii) fetching a matching template from said document-template database, wherein said matching template corresponds to said determined type of the identity document, and wherein said template includes one or more data fields; iii) overlaying said at least one two-dimensional image frame with said matching template, wherein each data field demarcates a respective data region of said overlaid image frame, to thereby obtain a geometrical correlation between said at least one two-dimensional image frame and said template; iv) selecting data regions, wherein said textual information is extracted from selected data regions; v) determining boundaries between text and background based on a minimum contrast between a character and other colored data on said image frame of the identity document, approximately within the boundaries of said selected data regions; vi) extracting black and white glyphs using at least a portion of said determined boundaries; and vii) applying said OCR routine on said black and white glyphs, thereby recognizing symbols and characters. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17)
-
-
18. A method for extracting information from an identity document, the method comprising the steps of:
-
a) providing a system for extracting information from an identity document including; i) at least one light source for illuminating the identity document; ii) at least one digital camera having a lens and a two-dimensional sensor array for acquiring at least one two-dimensional image frame of the identity document; and iii) a processor, being in communication flow with a document-template database, for processing said at least one two-dimensional image frame, b) determining a type of the identity document; c) fetching a matching template from said document-template database, wherein said matching template corresponds to said determined type of the identity document, and wherein said template includes one or more data fields; d) overlaying said at least one two-dimensional image frame with said matching template, wherein each data field demarcates a respective data region of said overlaid image frame, to thereby obtain a geometrical correlation between said at least one two-dimensional image frame and said template; e) selecting data regions, wherein the information is extracted from selected data regions; f) determining boundaries between text and background based on a minimum contrast between a character and other colored data on said image frame of the identity document, approximately within the boundaries of said selected data regions; g) extracting black and white glyphs using at least a portion of said determined boundaries; h) applying an optical character recognition (OCR) routine on said black and white glyphs, thereby recognizing symbols and characters; and i) sending said recognized symbols and characters as alphanumerical information, wherein said determining of said type of the identity document and said determining boundaries in the identity document are performed before applying said OCR routine on said black and white glyphs. - View Dependent Claims (19, 20)
-
Specification