License plate recognition system and method

• US 8,953,846 B2
• Filed: 05/21/2013
• Issued: 02/10/2015
• Est. Priority Date: 12/25/2012
• Status: Expired due to Fees

• Current Assignee: National Chiao Tung University (Government of The Republic of China)
• Original Assignee: National Chiao Tung University (Government of The Republic of China)
• Inventors: Wu, Bing-Fei, Chen, Yen-Lin, Huang, Hao-Yu, Lin, Hsin-Yu
• Primary Examiner(s): Abdi, Amara
• Application Number:
  US13/898,897
• Publication Number:
  US 20140177925A1
• Time in Patent Office: 630 Days
• Field of Search: None
• US Class Current: 382/105
• CPC Class Codes: G06V 20/625 License plates
• CPC Class Codes: G06V 30/413 Classification of content, e.g. text, photographs or tables
• CPC Class Codes: G08G 1/0175 by photographing vehicles, e.g. when violating traffic rules

Abstract

A license plate recognition system and method recognizes a license plate attached to a vehicle and bears a license plate number. The system includes a database module, a reducing module, an image capturing unit, a sorting unit, and a recognizing module. The database module stores license plate position sample type-related data and license plate number sample type-related data whose quantity is reducible by the reducing module. The image capturing unit captures license plate images at different imaging resolutions to generate license plate position image eigenvalue and license plate number image eigenvalue. The sorting unit computes the license plate position image eigenvalue to generate license plate position type-related data and computes the license plate number image eigenvalue to generate license plate number type-related data. The recognizing module compares the license plate number type-related data with the license plate number sample type-related data, thereby recognizing the license plate number.

Claims

1. A license plate recognition system for use in recognizing a license plate attached to a vehicle, the license plate bearing a license plate number, the license plate recognition system comprising:

a database module comprising a license plate position database and a license plate number database, the license plate position database storing a plurality of license plate position sample type-related data, and the license plate number database storing a plurality of license plate number sample type-related data;

a reducing module connected to the database module and adapted to reduce quantity of the license plate position sample type-related data and quantity of the license plate number sample type-related data in the database module;

an image capturing unit for capturing a license plate image at a first imaging resolution to compute the license plate image with a histogram of oriented gradients (HOG) algorithm and thereby generate an eigenvalue of license plate position image, and capturing a license plate image at a second imaging resolution to compute the license plate image with the HOG algorithm and thereby generate an eigenvalue of license plate number image, wherein the second imaging resolution is of a level higher than that of the first imaging resolution;

a sorting unit connected to the image capturing unit to compute an eigenvalue of the license plate position image with principal component analysis (PCA) and support vector machine (SVM) and thereby create license plate position type-related data, and compute an eigenvalue of the license plate number image with the PCA and the SVM and thereby create license plate number type-related data; and

a recognizing module connected to the sorting unit and the database module to compare the license plate position type-related data with the license plate position sample type-related data, compare the license plate number type-related data with the license plate number sample type-related data after determining that the license plate position type-related data is identical with one of the license plate position sample type-related data, and recognize the license plate number of the license plate by the license plate number sample type-related data identical with the license plate number type-related data after determining that the license plate number type-related data is identical with one of the license plate number sample type-related data.

2. The license plate recognition system of claim 1, further comprising an updating module connected to the database module and the image capturing unit, the updating module receiving a plurality of new license plate images from the image capturing unit and generating a plurality of new license plate image frames.

3. The license plate recognition system of claim 2, wherein the sorting unit is connected to the database module and uses the PCA and the SVM to compute an image of one of the new license plate image frames to thereby create new license plate position sample type-related data and new license plate number sample type-related data.

4. A license plate recognition method for use in recognizing a license plate attached to a vehicle, the license plate bearing a license plate number, the license plate recognition method comprising the steps of:

(a) storing a plurality of license plate position sample type-related data in a license plate position database and storing a plurality of license plate number sample type-related data in a license plate number database;

(b) capturing an image of the license plate at a first imaging resolution to thereby generate a first license plate image, and capturing an image of the license plate at a second imaging resolution to thereby generate a second license plate image;

(c) computing the first license plate image with a histogram of oriented gradients (HOG) algorithm to thereby generate an eigenvalue of license plate position image, and computing the second license plate image with the HOG algorithm to thereby generate an eigenvalue of license plate number image;

(d) computing the eigenvalue of the license plate position image and the eigenvalue of the license plate number image with principal component analysis (PCA) and support vector machine (SVM) to thereby create license plate position type-related data and license plate number type-related data; and

(e) comparing the license plate position type-related data with the license plate position sample type-related data, comparing the license plate number type-related data with the license plate number sample type-related data upon determination that the license plate position type-related data is identical with one of the license plate position sample type-related data, and recognizing the license plate number of the license plate by means of the license plate number sample type-related data identical with the license plate number type-related data upon determination that the license plate number type-related data is identical with one of the license plate number sample type-related data.

5. The license plate recognition method of claim 4, wherein step (a) is followed by step (f) of reducing quantity of the license plate position sample type-related data of the license plate position database and quantity of the license plate number sample type-related data of the license plate number database.

6. The license plate recognition method of claim 4, wherein step (b) further comprises the sub-step of, after failing to capture an image of the license plate at the first imaging resolution at a point of the vehicle, scanning continuously at the first imaging resolution at another point of the vehicle for an image of the license plate until an image of the license plate is captured from the vehicle.

7. The license plate recognition method of claim 6, wherein step (b) is followed by the steps of:

(g) calculating a total of marginal pixels of the first license plate image; and

(h) dividing the first license plate image into a plurality of image blocks, so as to determine a block similarity level between the image blocks,

wherein, upon determination that both the total of the pixels and the block similarity level are less than a first threshold, a position of the license plate at the vehicle is determined, using the first license plate image.

8. The license plate recognition method of claim 4, wherein the step (e) further comprises the sub-step of recognizing the license plate number of the license plate as a new license plate upon determination that the license plate position type-related data is not identical with any one of the license plate position sample type-related data.

9. The license plate recognition method of claim 8, wherein step (e) is followed by the steps of:

(i) capturing the new license plate images consecutively to thereby obtain a plurality of new license plate image frames; and

(j) computing the new license plate image frames to thereby create and store in the license plate position database new license plate position sample type-related data pertaining to the new license plate and to thereby create and store in the license plate number database new license plate number sample type-related data pertaining to the new license plate.

10. The license plate recognition method of claim 9, wherein the step (j) further comprises the sub-step of computing the new license plate image frames with the HOG algorithm, the PCA, and the SVM.

11. The license plate recognition method of claim 10, wherein the step (j) further comprises the sub-steps of:

(k) fetching a first image frame from the new license plate image frames and computing the new license plate image corresponding to the first image frame with the HOG algorithm to thereby generate a standard license plate image sample eigenvalue, wherein the standard license plate image sample eigenvalue correlates with the new license plate position sample type-related data and the new license plate number sample type-related data; and

(l) fetching a second image frame from the new license plate image frames, computing the new license plate image corresponding to the second image frame with the HOG algorithm to thereby generate an image sample eigenvalue, comparing the standard license plate image sample eigenvalue with the image sample eigenvalue to calculate a difference therebetween, replacing the standard license plate image sample eigenvalue with the image sample eigenvalue if the difference is less than the second threshold, keeping the standard license plate image sample eigenvalue unchanged if the difference is not less than the second threshold, storing the new license plate position sample type-related data in the license plate position database, and storing the new license plate number sample type-related data in the license plate number database.

12. The license plate recognition method of claim 9, wherein the new license plate image frames are in the number of 10 or more.