APPARATUS AND METHOD FOR RECOGNIZING IDENTIFIER OF VEHICLE

• US 20120057756A1
• Filed: 09/01/2011
• Published: 03/08/2012
• Est. Priority Date: 09/02/2010
• Status: Active Grant

• Current Assignee: Electronics and Telecommunications Research Institute
• Original Assignee: Electronics and Telecommunications Research Institute
• Inventors: YOON, Ho Sub, KIM, Do Hyung, LEE, Jae Yeon, KIM, Jae Hong, CHI, Su Young, SOHN, Joo Chan, BAN, Kyu Dae, YOON, Young Woo, YUN, Woo Han
• Granted Patent: US 8,548,201 B2
• Time in Patent Office: Days
• US Class Current: 382/104
• CPC Class Codes: G06F 18/214 Generating training patterns; Bootstrap methods, e.g. bagging or boosting
• CPC Class Codes: G06V 10/774 Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
• CPC Class Codes: G06V 20/63 Scene text, e.g. street names

Abstract

The present invention detects a candidate ROI group associated with character strings/figure strings on the basis of a result acquired through prior learning of various types of license plates, verifies the interested region candidate group detected by using at least one condition of five predetermined conditions, and determines an MBR region in the selected ROI region from the verified interested region candidate group by considering a ratio between the height and width of the ROI region to recognize the license plate for the automobile. According to the present invention, it is possible to automatically detect the location of the license plate regardless of various types of license plate specifications defined for each of countries.

Claims

1. A apparatus for recognizing an identifier of a vehicle, comprising:

an ROI detecting unit detecting ROIs from an input image on the basis of a prior learning result by images for training;

an identifier region selecting unit selecting the identifier region including the identifier of the vehicle from ROIs detected according to a predetermined criterion; and

an identifier recognizing unit analyzing the identifier region selected according to a type of the selected identifier region to recognize the vehicle identifier.

2. The apparatus of claim 1, further comprising:

a training image analyzing unit generating a common data region including common data among training images as the prior learning result by analyzing the training images which are different in at least one component of the content, location, and size of the identifier.

3. The apparatus of claim 2, wherein the training image analyzing unit includes:

a common data region defining portion comparing and analyzing the training images which are different in the components and defining the common data region including characters or figures;

a common data region extracting portion extracting the common data region defined for each of the training images;

a feature component extracting portion extracting a feature component from the extracted common data region; and

a prior learning result generating portion generating the prior learning result including successive character strings or successive figure strings on the basis of the extracted feature component.

4. The apparatus of claim 3, wherein the common data region extracting portion extracts common data regions defined as different sizes for each of the training images and thereafter, normalizes the sizes of the extracted common data regions to transmit the common data region extracted by the feature component extracting portion.

5. The apparatus of claim 3, wherein the training image analyzing unit further includes a background region processing portion extracting the background region without the common data region extracted for each of the training images and normalizing the size of the extracted background region.

6. The apparatus of claim 1, wherein the ROI detecting unit detects 2 to 5 ROIs for each input image.

7. The apparatus of claim 1, wherein the identifier region selecting unit includes:

a score allocating portion allocating a score to each of the detected ROIs according to a predetermined criterion; and

a region selecting portion selecting the identifier region from the detected ROIs on the basis of the allocated score.

8. The apparatus of claim 7, wherein the score allocating portion includes at least one of:

a first score allocating portion allocating the score to each of the detected ROIs by considering the relationship between a data region including data and a background region without data;

a second score allocating portion generating a histogram related to brightness for each detected ROI and allocating the score for each detected ROI according to a distance difference between peaks in the generated histogram; and

a third score allocating portion granting a weight value for each of features displaying the data region through the prior learning by the training images and reflecting the weight value to the extracted features for each detected ROI to allocate the score.

9. The apparatus of claim 8, wherein the first score allocating portion includes at least one of:

a first allocating portion allocating the score by calculating a region discriminating value between the data region and the background region;

a second allocating portion acquiring a first center point of the data region and a second center point of the background region and comparing a distance difference between a third center point of an ROI including the data region and the background region and the first center point and a distance difference between the third center point and the second center point with each other to allocate the score;

a third allocating portion acquiring a first average distance value between a first center pixel positioned at the center of the data region and at least one adjacent pixel positioned adjacent to the first center pixel and comparing the acquired first average distance value and the second average distance value previously determined through the prior learning by the training images with each other to allocate the score; and

a fourth allocating portion acquiring a third average distance value between a second center pixel positioned at the center of the background region and at least one adjacent pixel positioned adjacent to the second center pixel and comparing the acquired third average distance value and the second average distance value with each other to allocate the score.

10. The apparatus of claim 1, wherein the ROI detecting unit determines the size of an ROI to be detected by considering a ratio of a predetermined width value and a predetermined height value at the time of detecting the ROI.

11. The apparatus of claim 7, wherein the region selecting portion includes:

a temporary identifier region detecting portion detecting a temporary identifier region depending on whether data regions having the same size are sequentially present by performing labeling for each detected ROI on the basis of the allocated score or detecting the temporary identifier region depending on whether all successive data are included on the basis of a brightness value of the background region for each detected ROI on the basis of the allocated score; and

a region determining portion determining the identifier region depending on whether or not the detected temporary identifier region coincides with a predetermined criterion.

12. The apparatus of claim 11, further comprising:

an image inverting unit inverting the input image when the identifier region which coincides with the predetermined criterion is not selected.

13. The apparatus of claim 12, further comprising:

an identifier region generating unit combining the temporary identifier region selected by the temporary identifier region detecting portion and the temporary identifier region acquired by signal-processing the input image with each other to generate the identifier region when the identifier region which coincides with the predetermined criterion is not selected from the inverted image.

14. The apparatus of claim 1, further comprising:

an identifier region type determining unit determining the type of the selected identifier region by considering a ratio between a width value and a length value of the selected identifier region.

15. The apparatus of claim 1, wherein the identifier recognizing unit includes:

a component identifying portion identifying data positioned in the identifier region selected according to the type of the selected identifier region into at least two types of components; and

a vehicle identifier recognizing portion recognizing the vehicle identifier by using the identified components.

16. The apparatus of claim 15, wherein the component identifying portion identifies the data into a character string component and a figure string component, and

the vehicle identifier recognizing portion compares each of the identified components with previously stored components to recognize the vehicle identifier.

17. A method for recognizing an identifier of a vehicle, comprising:

detecting ROIs from an input image on the basis of a prior learning result by images for training;

selecting the identifier region including the identifier of the vehicle from ROIs detected according to a predetermined criterion; and

analyzing the identifier region selected according to a type of the selected identifier region to recognize the vehicle identifier.

18. The method of claim 17, further comprising:

generating a common data region including common data among training images as the prior learning result by analyzing the training images which are different in at least one component of the content, location, and size of the vehicle identifier.

19. The method of claim 17, wherein the selecting of the identifier region includes:

allocating a score to each of the detected ROIs according to a predetermined criterion; and

selecting the identifier region from the detected ROIs on the basis of the allocated score.

20. The method of claim 17, wherein the recognizing of the identifier includes:

identifying data positioned in the identifier region selected according to the type of the selected identifier region into at least two types of components; and

recognizing the vehicle identifier by using the identified components.