METHOD AND APPARATUS FOR EXTRACTING TEXT AREA, AND AUTOMATIC RECOGNITION SYSTEM OF NUMBER PLATE USING THE SAME

• US 20120148101A1
• Filed: 12/13/2011
• Published: 06/14/2012
• Est. Priority Date: 12/14/2010
• Status: Abandoned Application

• Current Assignee: Electronics and Telecommunications Research Institute
• Original Assignee: Electronics and Telecommunications Research Institute
• Inventors: YOON, Young Woo, Yoon, Ho Sub, Lee, Jae Yeon, Kim, Do Hyung, Chi, Su Young, Kim, Jae Hong, Sohn, Joo Chan, Ban, Kyu Dae
• Application Number:
  US13/325,035
• Publication Number:
  US 20120148101A1
• Time in Patent Office: Days
• US Class Current: 382/103
• CPC Class Codes: G06V 20/62 Text, e.g. of license plates, overlay texts or captions on TV images
• CPC Class Codes: G06V 20/625 License plates

Abstract

Disclosed is a method of extracting a text area, the method including generating a text area prediction value within an input second image based on a plurality of text area data stored in a database including geometric information about a text area of a first image, generating a text recognition result value by determining whether a text is recognized with respect to a probable text area within the input second image, and selecting a text area within the second image by combining the generated text area prediction value and text recognition result value.

Claims

1. A method of extracting a text area, comprising:

generating a text area prediction value within an input second image based on a plurality of text area data stored in a database including geometric information about a text area of a first image;

generating a text recognition result value by determining whether a text is recognized with respect to a probable text area within the input second image; and

selecting a text area within the second image by combining the generated text area prediction value and text recognition result value.

2. The method of claim 1, wherein:

the geometric information includes position and size information of the text area, and

the generating of the text area prediction value generates the prediction value based on similarity with N text area data stored in the database including the position and size information of the text area, N indicating a positive integer equal to or greater than 1.

3. The method of claim 2, wherein the text is meaningful visual information including at least one of a character, a number, a symbol, and a sign.

4. The method of claim 3, wherein the position and size information about the text area of the first image pre-stored in the database and the generated text recognition result value are learning information that is repeatedly used to select the text area within the second image.

5. The method of claim 2, wherein the database includes the position and size information about the text area of the first image in a form of numerical value information that is converted into a vector format.

6. The method of claim 5, wherein the vector format is a format that includes an absolute value with respect to the text area or a positional relative value with another text area.

7. The method of claim 6, wherein the generating of the text area prediction value further comprises:

generating a missing value estimate by predicting a missing value of the text area based on the database and text extraction information from the second image; and

generating a first score map storing an estimation probability about the missing value estimate based on all of the predicted missing value estimates.

8. The method of claim 7, wherein:

the generating of the text recognition result value recognizes whether the text exists with respect to all areas within the second image, and

an absolute value or a relative value with respect to the text area includes all of the horizontal and vertical position values within the second image and includes minimum to maximum sizes of a width and a height of the text area.

9. The method of claim 8, wherein the generating of the text recognition result value further comprises:

generating a second score map storing an estimation probability of whether the recognized text exists.

10. The method of claim 9, wherein the selecting of the text area further comprises:

generating a third score map merged by adding the same standard of the generated first score map and second score map, and

the selecting of the text area selects a text area having the highest score in the generated third score map.

11. The method of claim 10, wherein the selecting of the text area excludes the text area having the highest score in the generated third score map from selectable text area candidates when the text area having the highest score in the generated third score map overlaps an area selected as another text area by at least a predetermined range.

12. The method of claim 1, after the selecting of the text area, further comprising:

determining whether a text area extraction operation within the second image is completed by repeatedly performing the text area extraction method.

13. The method of claim 12, wherein the second image is an image of a notice plate, and

the determining of whether the text area extraction operation is completed compares the number of the extracted text areas according to a notice plate indication standard of each country.

14. An apparatus for extracting a text area, comprising:

a database to include geometric information about a text area of a first image;

a missing value predicting unit to generate a missing value estimate by predicting a missing value of a text area within an input second image based on a plurality of text area data stored in the database;

a first score map generating unit to generate a first score map storing an estimation probability about the missing value estimate based on the predicted missing value estimate;

a text recognition unit to generate a text recognition result value by determining whether a text is recognized with respect to a probable text area within the input second image;

a second score map generating unit to generate a second score map storing an estimation probability of whether the recognized text exists; and

a text area selecting unit to select the text area within the second image by combining the generated first score map and second score map.

15. An automatic number plate recognition system, comprising:

a number plate detecting unit to detect a number plate image from an external image photographed using a camera;

a text area extracting unit to generate a text area prediction value within the detected number plate image based on a database including geometric information about a text area within a pre-stored number plate image, to generate a text recognition result value by determining whether a text is recognized with respect to a probable text area within the number plate image, and to select a text area within the number plate image by combining the generated text area prediction value and text recognition result value; and

a text identifying unit to identify a text indicated within the extracted text area.