System for template based extracting information from an identity card

• US 8,610,966 B2
• Filed: 02/03/2011
• Issued: 12/17/2013
• Est. Priority Date: 10/06/2004
• Status: Active Grant

• Current Assignee: Acuant Inc. (GB Group Plc)
• Original Assignee: Card Scanning Solutions Inc. (GB Group Plc)
• Inventors: Hatzav, Iuval, Ben-Shahar, Ilan, Sharaby, Giyora, Hatzav, Daniel
• Primary Examiner(s): Worku, Negussie
• Application Number:
  US13/020,058
• Publication Number:
  US 20110128360A1
• Time in Patent Office: 1,048 Days
• Field of Search: 358/474, 358/475, 358/473, 358/462, 358/483, 358/909.1, 358/509, 382/181, 348/61, 348/E07.085, 348/156, 348/96, 348/207.1
• US Class Current: 358/474
• CPC Class Codes: G03B 15/00 Special procedures for taking photographs; Apparatus therefor
• CPC Class Codes: G03B 17/00 Details of cameras or camera bodies; Accessories therefor lens hoods or caps G03B11/04
• CPC Class Codes: G03B 27/323 Copying cameras G03B27/34 - G03B27/70 take precedence
• CPC Class Codes: G06V 10/143 Sensing or illuminating at different wavelengths
• CPC Class Codes: G06V 30/416 Extracting the logical structure, e.g. chapters, sections or page numbers; Identifying elements of the document, e.g. authors

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.

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.

2. A system according to claim 1, wherein said OCR routine further includes:

viii) sending said recognized symbols and characters as alphanumerical information.

3. A system according to claim 1, the system further comprising:

d) a box for housing said digital camera and said light source; and

a communication link for connecting said box to said processor.

4. A system according to claim 1, the system further comprising:

d) mirrors for simultaneously imaging both sides of said document.

5. A system according to claim 1, wherein said at least one light source includes at least two light sources for respectively illuminating both sides of said document;

and wherein said at least one digital camera includes two digital cameras for respectively imaging both sides of the identity document.

6. A system according to claim 1, wherein said light source includes a light diffuser for even illumination of the identity document.

7. A system according to claim 6, wherein said light diffuser is a light reflector.

8. A system according to claim 7, the system further comprising:

d) a light baffle for preventing directly illuminating the identity document by said light source.

9. A system according to claim 1, wherein extraction by said OCR routine is facilitated by prior knowledge of said type of the identity document.

10. A system according to claim 1, wherein acquisition of said at least one two-dimensional image frame is automatically initiated by sensing proper placement of the identity document.

11. A system according to claim 10, the system further comprising:

d) a card sensor for facilitating said sensing proper placement of the identity document.

12. A system according to claim 10, wherein said sensing proper placement of the identity document is performed by acquiring image frames and performing at least partial image processing of said image frames.

13. A system according to claim 12, wherein said image frames used for sensing proper placement of the identity document are of lower quality than image frames used for extracting information.

14. A system according to claim 1, wherein said alphanumerical information is used to save a user from manually inputting said alphanumerical information.

15. A system according to claim 1, wherein said textual information is used to augment a database.

16. A system according to claim 1, wherein said textual information is used as key words to perform a search.

17. A system according to claim 1, wherein said processor is configured to perform pre-processing steps, for facilitating better-quality image frames of the identity document, before applying said OCR routine for extracting textual information from the identity document in the form of alphanumerical information.

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.

19. A method according to claim 18, the method further comprising:

j) pre-processing said at least one two-dimensional image frame, for facilitating better-quality image frames of the identity document, wherein said pre-processing is performed before applying said OCR routine on said black and white glyphs.

20. A method according to claim 19, wherein said pre-processing includes at least one of the following steps selected from the group consisting of:

i. determining said type of the identity document and removing an image-frame background from the identity document;

ii. calibrating illumination parameters before acquiring an image frame of the identity document for extracting the information from the identity document; and

iii. correcting image distortions caused by known lens distortions.