Apparatus and method for recognizing traffic sign board

• US 9,501,703 B2
• Filed: 07/15/2014
• Issued: 11/22/2016
• Est. Priority Date: 01/06/2014
• Status: Active Grant

• Current Assignee: Hyundai Mobis Co., Ltd.
• Original Assignee: Hyundai Mobis Co., Ltd.
• Inventors: Oh, Huen, Byun, Hye Ran, Lim, Kwang Yong, Lee, Tae Woo
• Primary Examiner(s): ENTEZARI, MICHELLE M
• Application Number:
  US14/332,084
• Publication Number:
  US 20150193663A1
• Time in Patent Office: 861 Days
• Field of Search: None
• US Class Current: 1/1
• CPC Class Codes: G06F 18/2148 characterised by the process organisation or structure, e.g. boosting cascade
• CPC Class Codes: G06V 10/36 Applying a local operator, i.e. means to operate on image points situated in the vicinity of a given point; Non-linear local filtering operations, e.g. median filtering
• CPC Class Codes: G06V 10/467 Encoded features or binary features, e.g. local binary patterns [LBP]
• CPC Class Codes: G06V 10/7747 Organisation of the process, e.g. bagging or boosting
• CPC Class Codes: G06V 20/582 of traffic signs

Abstract

A method and apparatus for detecting and recognizing a traffic sign using a modified census transform (MCT) feature are disclosed. The traffic sign recognizing method according to an exemplary embodiment of the present invention includes detecting a traffic sign candidate region from an input image using a modified census transform (MCT) feature; verifying whether the candidate region corresponds to a traffic sign using the MCT feature histogram for the candidate region; and lassifying a region of interest into the corresponding traffic sign step by step using the MCT feature histogram for the verified candidate region.

Claims

1. A traffic sign recognizing method, comprising:

detecting a traffic sign candidate region from an input image using a modified census transform (MCT) feature;

verifying whether the candidate region corresponds to a traffic sign using the MCT feature histogram for the candidate region;

classifying a region of interest into the corresponding traffic sign using the MCT feature histogram for the verified candidate region; and

matching the region of interest using a radial symmetry detection (RSD) based on voting in a gradient direction using a deviation of an X axis value and a deviation of a Y axis value of each pixel of the input image.

2. The method of claim 1, wherein the detecting of a candidate region comprises:

designating a kernel window having a size with respect to a pixel to be transformed in the input image; and

comparing an average of pixel values of pixels included in the kernel window with individual pixel values to calculate n digit bits of MCT feature by assigning a value of 1 to the pixel, in response to the pixel value of the corresponding pixel being larger than the average of the pixel values, and assigning a value of 0 to the pixel, in response to the pixel value of the corresponding pixel being smaller than the average.

3. The method of claim 1, wherein the detecting of a candidate region comprises:

extracting the MCT feature for every pixel of the input image using the pixel value of the input image; and

classifying the candidate region from the input image using a classifier in which an MCT feature for the traffic sign is trained using at least one of an AdaBoost algorithm or a cascade algorithm.

4. The method of claim 1, wherein the detecting of the candidate region comprises:

extracting the MCT feature for every pixel of the input image using the pixel value of the input image;

classifying n features from the extracted MCT features using a classifier in which the MCT feature for the traffic sign is trained; and

classifying m features, excluding the n features, among remaining MCT features.

5. The method of claim 1, wherein the verifying comprises:

creating an MCT feature histogram for an MCT feature value and a frequency of the value using the MCT feature for each pixel of the candidate region; and

verifying the MCT feature histogram of the input image using a classifier in which the MCT feature histogram for the traffic sign is trained.

6. The method of claim 1, further comprising creating an MCT feature histogram for each of a plurality of regions within the region of interest to form a multilevel classification tree in which each level contained therein is used to classify the region of interest according to a type of traffic sign classification.

7. The method of claim 1, further comprising creating a first MCT feature histogram in a first region of the region of interest in an N-th level and a second MCT feature histogram in a second region of the region of interest in an N+1-th level.

8. The method of claim 1, further comprising:

classifying the traffic sign as a speed sign, in response to classifying the region of interest as a speed sign region, and classifying the traffic sign as a non-recognized target sign, in response to classifying the region of interest as an other sign region.

16. The method of claim 2, further comprising calculating bits resulting in obtaining a decimal feature vector, based on binary-coded result values for pixels not to be transformed.

17. The method of claim 1, further comprising creating a first MCT feature histogram for a first half of the region of interest and a second MCT feature histogram for a second half of the region of interest.

18. The method of claim 1, wherein the classifying the region of interest into the corresponding traffic sign is performed step-by-step.

19. The method of claim 1, wherein the matching further comprises determining a point having a highest voted value as a center point of a circle.

20. The method of claim 19, wherein the matching further comprises matching the region of interest with a rectangular region encompassing a radius of the circle, the radius is obtained in response to the center point of the circle being considered as a center of the region of interest, and the radius is one half of a length corresponding to one edge of the region of interest.

21. A non-transitory computer-readable medium storing instructions that, when executed by one or more processors, cause the one or more processors to perform the method of claim 1.

9. A traffic sign recognizing apparatus, comprising:

a processor configured to;

detect a traffic sign candidate region from an input image using a modified census transform (MCT) feature;

verify whether the candidate region corresponds to a traffic sign using the MCT feature histogram for the candidate region;

classify a region of interest into the corresponding traffic sign using the MCT feature histogram for the verified candidate region; and

match the region of interest using a radial symmetry detection (RSD) based on voting in a gradient direction using a deviation of an X axis value and a deviation of a Y axis value of each pixel of the input image.

10. The apparatus of claim 9, wherein the processor is further configured to:

designate a kernel window having a size with respect to a pixel to be transformed in the input image; and

compare an average of pixel values of pixels included in the kernel window with individual pixel values to calculate n digit bits of MCT feature by assigning a value of 1 to the pixel, in response to the pixel value of the corresponding pixel being larger than the average of the pixel values, and assigning a value of 0 to the pixel, in response to the pixel value of the corresponding pixel being smaller than the average.

11. The apparatus of claim 9, wherein the processor is further configured to:

extract the MCT feature for every pixel of the input image using the pixel value of the input image; and

classify the candidate region from the input image using a classifier in which an MCT feature for the traffic sign is trained using at least one of an AdaBoost algorithm and a cascade algorithm.

12. The apparatus of claim 9, wherein the processor is further configured to:

extract the MCT feature for every pixel of the input image using the pixel value of the input image;

classify n features from the extracted MCT features using a classifier in which the MCT feature for the traffic sign is trained as a first step; and

classify m features, excluding the n features, among remaining MCT features.

13. The apparatus of claim 9, wherein the processor is further configured to:

create an MCT feature histogram for an MCT feature value and a frequency of the value using the MCT feature for each pixel of the candidate region; and

verify the MCT feature histogram of the input image using a classifier in which the MCT feature histogram for the traffic sign is trained.

14. The apparatus of claim 9, wherein the processor is further configured to create an MCT feature histogram for each of a plurality of regions within the region of interest to form a multilevel classification tree in which each level contained therein is used to classify the region of interest according to a type of traffic sign classification.

15. The apparatus of claim 14, wherein the processor is further configured to create a first MCT feature histogram in a first region of the region of interest in an N-th level and a second MCT feature histogram in a second region of the region of interest in an N+1-th level.