System and method for recognizing vehicle license plate information

• US 10,311,319 B2
• Filed: 05/18/2017
• Issued: 06/04/2019
• Est. Priority Date: 10/31/2016
• Status: Active Grant

• Current Assignee: Electronics and Telecommunications Research Institute
• Original Assignee: Electronics and Telecommunications Research Institute
• Inventors: Kim, Pyong Kun, Kang, Hyun Woo, Kim, Kwang Ju, Lee, Jong Taek, Lim, Kil Taek, Chung, Yun Su, Han, Byung Gil
• Primary Examiner(s): Abdi, Amara
• Application Number:
  US15/598,414
• Publication Number:
  US 20180121744A1
• Time in Patent Office: 747 Days
• CPC Class Codes: G06V 10/993 Evaluation of the quality of the acquired pattern
• 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

Provided is a technology for recognizing vehicle license plate information which includes detecting a position of a license plate from a vehicle image obtained by imaging a vehicle, extracting individual character images by separating a plurality of characters from a region of the detected license plate in the vehicle image and extracting a license plate type of the license plate according to a predetermined criterion, recognizing a license plate character string on the basis of the extracted individual character images and the license plate type and outputting the recognized license plate character string.

Claims

1. A vehicle license plate information recognition system comprising:

a license plate detecting module configured to detect a position of a license plate from a vehicle image obtained by imaging a vehicle;

a character separation module configured to separate a plurality of characters from a region of the detected license plate in the vehicle image, extract individual character images, and extract a license plate type of the license plate according to a predetermined criterion;

a character recognition module configured to recognize a license plate character string on the basis of the extracted individual character images and the license plate type; and

a license plate information output module configured to output the recognized license plate character string,

wherein the character recognition module recognizes a usage character in the individual character images through type-specific character recognition and universal character recognition, which each employ different classification criteria, the usage character corresponding to at least one of a time at which the license plate was generated, a location at which the vehicle is registered, and a usage for which the vehicle is used,

wherein the type-specific character recognition recognizes the usage character using both an individual character image and type information regarding the license plate type,

wherein the universal character recognition recognizes the usage character using the individual character image without using the type information,

wherein the character recognition module compares a result of the type-specific character recognition and the universal character recognition, and

wherein the character recognition module performs, based on a result of the comparison, one of a process of generating a license plate character string, a process of reexamining a license plate type with respect to an individual character image, and a process of rejecting a result of a character recognition with respect to an individual character image.

2. The system of claim 1, wherein the character recognition module comprises:

a type-specific character recognizer configured to recognize a type-specific usage character on the basis of the individual character image of the license plate and the license plate type information; and

an universal character recognizer configured to recognize a usage character on the basis of the individual character image, wherein the usage character is recognized by employing a classification criterion that is different from a classification criterion of the type-specific character recognizer and is a high level criterion including the classification criterion of the type-specific character recognizer.

3. The system of claim 2, wherein:

the character recognition module is configured to, when results of the usage character recognition in the individual character images of the license plate by the type-specific character recognizer and the universal character recognizer are identical to each other;

generate and output the license plate character string in a case in which results of recognition of the remaining individual character images except for the usage character are identical to each other or similar to each other with a similarity of a predetermined threshold value or more; and

reject results of recognition of the individual character image in a case in which the results of recognition of the remaining individual character images have a similarity less than the predetermined threshold value.

4. The system of claim 2, wherein:

the character recognition module is configured to, when results of the usage character recognition in the individual character images of the license plate by the type-specific character recognizer and the universal character recognizer are different from each other;

re-perform a character recognition by reexamining the license plate type in a case in which the results of the usage character recognition have a similarity of a predetermined threshold value or more; and

reject a result of the individual character image in a case in which the results of the usage character recognition have a similarity less than the predetermined threshold value.

5. The system of claim 1, wherein the classification criteria for the license plate type includes at least one of a size of a license plate, a type of a usage character, a combination of a background color and a character color, and a difference between background brightness and character brightness.

6. A method for recognizing vehicle license plate information through a vehicle license plate information recognition system, the method comprising:

detecting a position of a license plate from a vehicle image obtained by imaging a vehicle;

separating a plurality of characters from a region of the detected license plate in the vehicle image, extracting individual character images, and extracting a license plate type of the license plate according to a predetermined criterion;

recognizing a license plate character string on the basis of the extracted individual character images and the license plate type; and

outputting the recognized license plate character string,

wherein the recognizing of the license plate character string comprises recognizing a usage character in the individual character images through type-specific character recognition and universal character recognition, which each employ different classification criteria, the usage character corresponding to at least one of a time at which the license plate was generated, a location at which the vehicle is registered, and a usage for which the vehicle is used,

wherein recognizing the usage character through the type-specific character recognition includes using both an individual character image and type information regarding the license plate type,

wherein recognizing the usage character through the universal character recognition includes using the individual character image without using the type information, and

wherein the recognizing of the license plate character string further comprises comparing a result of the type-specific character recognition and the universal character recognition and performing, based on a result of the comparison, one of a process of generating a license plate character string, a process of reexamining a license plate type with respect to an individual character image, and a process of rejecting a result of a character recognition with respect to an individual character image.

7. The method of claim 6, wherein the recognizing of the license plate character string comprises:

performing the type-specific character recognition in which a type-specific usage character is recognized on the basis of the individual character image of the license plate and the license plate type information;

performing a usage character recognition of recognizing a usage character on the basis of the individual character image, wherein the usage character is recognized by employing a classification criterion that is different from a classification criterion of the type-specific character recognition and is a high level criterion including the classification criterion of the type-specific character recognition; and

performing one of the process of generating a license plate character string, the process of reexamining a license plate, and the process of rejecting a character recognition result on the basis of the result of comparing the usage characters recognized through the type-specific character recognition and the universal character recognition with each other.

8. The method of claim 7, wherein the recognizing of the license plate character string comprises, when results of the usage character recognition in the individual character images of the license plate by the type-specific character recognition and the universal character recognition are identical to each other,

generating and outputting a license plate character string in a case in which results of recognition of the remaining individual character images except for the usage character are identical to each other or similar to each other with a similarity of a predetermined threshold value or more, and

rejecting results of recognition of the individual character image in a case in which the results of recognition of the remaining individual character images have a similarity less than the predetermined threshold value.

9. The method of claim 7, wherein the recognizing of the license plate character string comprises, when results of the usage character recognition in the individual character images of the license plate by the type-specific character recognition and the universal character recognition are different from each other,

reexamining the license plate type and re-performing a character recognition in a case in which the results of the usage character recognition have a similarity of a predetermined threshold value or more, and

rejecting a result of recognition of the individual character image in a case in which the results of the usage character recognition have a similarity less than the predetermined threshold value.

10. The method of claim 6, wherein the classification criteria of the license plate type includes at least one of a size of a license plate, a type of a usage character, a combination of a background color and a character color, and a difference between background brightness and character brightness.

11. A vehicle license plate information recognition system comprising:

a type-specific character recognizer configured to recognize a type-specific usage character on the basis of an individual character image of a license plate in an image obtained by imaging a vehicle and a license plate type of the license plate; and

a universal character recognizer configured to recognize a usage character on the basis of the individual character image, wherein the usage character is recognized by employing a classification criterion that is different from a classification criterion of the type-specific character recognizer and is a high level criterion including the classification criterion of the type-specific character recognizer,

wherein one of a process of generating a license plate character string, a process of reexamining a license plate type with respect to an individual character image, and a process of rejecting a result of character recognition with respect to an individual character image is performed on the basis of a result of comparing usage characters recognized through the type-specific character recognizer and the universal character recognizer with each other.

12. The system of claim 11, wherein, when results of the usage character recognition in the individual character images of the license plate by the type-specific character recognizer and the universal character recognizer are identical to each other,

a license plate character string is generated and output in a case in which results of recognition of the remaining individual character images except for the usage character are identical to each other or similar to each other with a similarity of a predetermined threshold value or more, and

a result of recognition of the individual character image is rejected in a case in which the results of recognition of the remaining individual character images have a similarity less than the predetermined threshold value.

13. The system of claim 11, wherein, when results of the usage character recognition in the individual character images of the license plate by the type-specific character recognizer and the universal character recognizer are different from each other,

character recognition is re-performed by reexamining the license plate type in a case in which the results of the usage character recognition have a similarity of a predetermined threshold value or more, and

a result of recognition of the individual character image is rejected in a case in which the results of the usage character recognition have a similarity less than the predetermined threshold value.