TRAFFIC LIGHT RECOGNITION SYSTEM AND METHOD

• US 20200134333A1
• Filed: 09/03/2019
• Published: 04/30/2020
• Est. Priority Date: 10/31/2018
• Status: Active Grant

• Current Assignee: Cognizant Technology Solutions India Private Limited (Cognizant Technology Solutions Corp.)
• Original Assignee: Cognizant Technology Solutions India Private Limited (Cognizant Technology Solutions Corp.)
• Inventors: Vishal, Kumar, Srinivasa, Arvind Channarayapatna, Mishra, Ritesh, Gundimeda, Venugopal
• Granted Patent: US 11,087,153 B2
• Time in Patent Office: Days
• CPC Class Codes: G06F 18/2115 by evaluating different subsets according to an optimisation criterion, e.g. class separability, forward selection or backward elimination
• CPC Class Codes: G06F 18/2451 linear, e.g. hyperplane
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 2207/10024 Color image
• CPC Class Codes: G06T 7/75 involving models
• CPC Class Codes: G06T 7/90 Determination of colour characteristics
• CPC Class Codes: G06V 10/764 using classification, e.g. of video objects
• CPC Class Codes: G06V 10/771 Feature selection, e.g. selecting representative features from a multi-dimensional feature space
• CPC Class Codes: G06V 20/584 of vehicle lights or traffic lights

Abstract

The present disclosure is directed to a traffic light recognition system and method for advanced driver assistance systems (ADAS) and robust to variations in illumination, partial occlusion, climate, shape and angle at which traffic light is viewed. The solution performs a real time recognition of traffic light by detecting the region of interest, where extracting the region of interest is achieved by projecting the sequence of frames into a kernel space, binarizing the linearly separated sequence of frames, identifying and classifying the region of interest as a candidate representative of traffic light. With the aforesaid combination of techniques used, traffic light can be conveniently recognized from amidst closely similar appearing objects such as vehicle headlights, tail or rear lights, lamp posts, reflections, street lights etc. with enhanced accuracy in real time.

Claims

1. ) A traffic light recognition system, comprising:

a kernelization module configured to;

receive a sequence of frames captured by an imaging device; and

project the sequence of frames into a kernel space;

a binarization module configured to binarize, based on a dynamically determined threshold, the sequence of frames separated in the kernel space;

a decision tree module configured to identify a set of candidate blobs in the binarized sequence of frames based on a set of predefined features; and

a classification module configured to determine if the identified set of candidate blobs is a candidate for representing a traffic light.

2. ) The traffic light recognition system, according to claim 1, further comprising a search area minimization module configured to:

receive the plurality of images captured of a region external to the imaging device;

analyze the received plurality of images to determine a search area having a positional relationship therewith; and

set a minimized search area for transmitting an optimized search area to the kernelization module.

3. ) The traffic light recognition system, according to claim 1, wherein the kernelization module is configured to make linearly separable the sequence of frames comprising at least one region of interest from environment thereof by a hyper plane.

4. ) The traffic light recognition system, according to claim 3, wherein the region of interest includes traffic light, vehicle head lights, tail lights, street lights, pedestrian lights, vehicular parking lights, road signs or a combination thereof.

5. ) The traffic light recognition system, according to claim 3, wherein the kernelization module is configured to make the detected region of interest linearly separable from the environment thereof based on characteristic properties of the detected at least one region of interest determined to be above dynamically determined threshold values.

6. ) The traffic light recognition system, according to claim 5, wherein the characteristic properties of the detected at least one region of interest includes luminance and saturation properties of plurality of pixels within the detected region of interest.

7. ) The traffic light recognition system, according to claim 1, wherein the kernelization module is further configured to normalize the kernel space to a monochrome space.

8. ) The traffic light recognition system, according to claim 7, wherein the binarization module is configured to convert monochrome image separated in the monochrome space into binarize image by retaining top 1 percentile as the dynamically determined threshold.

9. ) The traffic light recognition system, according to claim 1, wherein the binarization module is configured to binarize the sequence of frames separated in the kernel space by retaining top 1 percentile as the dynamically determined threshold.

10. ) The traffic light recognition system, according to claim 1, wherein the decision tree module is configured to perform first level of false positive elimination, identify the set of candidate blobs within the binarized sequence of frames based on the set of predefined features that corresponds to geometric parameters as well as identified vehicle regions.

11. ) The traffic light recognition system, according to claim 10, some of the predefined features that corresponds to geometric parameters comprising a minimum and/or maximum size bound of blobs with respect to image resolution, height, width, shape, width to height ratio and other mensuration parameters.

12. ) The traffic light recognition system, according to claim 10, wherein the classification module is further configured to perform second level of false positive elimination by mapping the identified set of candidate blobs back to the original sequence of frames and determining if the mapped set of candidate blobs is the candidate for representing the traffic light.

13. ) A traffic light recognition method, comprising:

receiving a sequence of frames captured by an imaging device;

projecting the sequence of frames into a kernel space;

binarizing, based on a dynamically determined threshold, the sequence of frames separated in the kernel space;

identifying a set of candidate blobs in the binarized sequence of frames based on a set of predefined features; and

determining if the identified set of candidate blobs is a candidate for representing a traffic light.

14. ) The traffic light recognition method, according to claim 13, further comprising:

receiving the sequence of frames captured of a region external to the imaging device;

analyzing the received sequence of frames to determine a search area having a positional relationship therewith; and

setting a minimized search area for transmitting an optimized search area for kernelization.

15. ) The traffic light recognition method, according to claim 13, further comprising making linearly separable the sequence of frames comprising at least one region of interest from environment thereof by a hyper plane.

16. ) The traffic light recognition method, according to claim 15, wherein the region of interest includes traffic light, vehicle head lights, tail lights, street lights, pedestrian lights, vehicular parking lights, road signs or a combination thereof.

17. ) The traffic light recognition method, according to claim 15, wherein the detected region of interest is made linearly separable from the environment thereof based on characteristic properties of the detected at least one region of interest determined to be above dynamically determined threshold values.

18. ) The traffic light recognition method, according to claim 17, wherein the characteristic properties of the detected at least one region of interest includes luminance and saturation properties of plurality of pixels within the detected region of interest.

19. ) The traffic light recognition method, according to claim 13, further comprising normalizing the kernel space to a monochrome space.

20. ) The traffic light recognition method, according to claim 19, further comprising converting monochrome image separated in the monochrome space into binarize image by retaining top 1 percentile as the dynamically determined threshold.

21. ) The traffic light recognition method, according to claim 13, further comprising binarizing the sequence of frames separated in the kernel space by retaining top 1 percentile as the dynamically determined threshold.

22. ) The traffic light recognition method, according to claim 13, further performing first level of false positive elimination, identifying the set of candidate blobs within the binarized sequence of frames based on the set of predefined features that corresponds to geometric parameters as well as identified vehicle regions.

23. ) The traffic light recognition method, according to claim 22, some of the predefined features that corresponds to geometric parameters comprising a minimum and/or maximum size bound of blobs with respect to image resolution, height, width, shape, width to height ratio and other mensuration parameters.

24. ) The traffic light recognition method, according to claim 22, further performing second level of false positive elimination by mapping the identified set of candidate blobs back to the original sequence of frames and determining if the mapped set of candidate blobs is the candidate for representing the traffic light.