SHORT-TIME STOPPING DETECTION FROM RED LIGHT CAMERA VIDEOS

• US 20150332097A1
• Filed: 05/15/2014
• Published: 11/19/2015
• Est. Priority Date: 05/15/2014
• Status: Active Grant

• Current Assignee: Conduent Business Services, LLC (Conduent, Inc.)
• Original Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Inventors: Bulan, Orhan, Burry, Aaron M., Loce, Robert P.
• Granted Patent: US 9,275,286 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 18/24 Classification techniques
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 2207/10024 Color image
• CPC Class Codes: G06T 2207/30232 Surveillance
• CPC Class Codes: G06T 2207/30236 Traffic on road, railway or crossing
• CPC Class Codes: G06T 2207/30241 Trajectory
• CPC Class Codes: G06V 20/52 Surveillance or monitoring of activities, e.g. for recognising suspicious objects recognising microscopic objects G06V20/69
• CPC Class Codes: G06V 20/54 of traffic, e.g. cars on the road, trains or boats
• CPC Class Codes: G08G 1/0175 by photographing vehicles, e.g. when violating traffic rules

Abstract

A method for detecting a vehicle running a stop signal positioned at an intersection includes acquiring a sequence of frames from at least one video camera monitoring an intersection being signaled by the stop signal. The method includes defining a first region of interest (ROI) including a road region located before the intersection on the image plane. The method includes searching the first ROI for a candidate violating vehicle. In response to detecting the candidate violating vehicle, the method includes tracking at least one trajectory of the detected candidate violating vehicle across a number of frames. The method includes classifying the candidate violating vehicle as belonging to one of a violating vehicle and a non-violating vehicle based on the at least one trajectory.

Claims

1. A method to detect a vehicle running a stop signal positioned at an intersection, the method comprising:

acquiring a sequence of frames from at least one video camera monitoring an intersection being signaled by the stop signal;

defining a first region of interest (ROI) including a road region located before the intersection on the image plane;

searching the first ROI for a candidate violating vehicle;

in response to detecting the candidate violating vehicle, tracking at least one trajectory of the detected candidate violating vehicle across a number of frames;

classifying the candidate violating vehicle as belonging to one of a violating vehicle and a non-violating vehicle based on the at least one trajectory.

2. The method of claim 1 further comprising:

in response to the stop signal being a traffic light, determining a color of a signal lamp displayed on the traffic light as belonging to one of red, yellow, and green; and

,

classifying the candidate violating vehicle as belonging to one of a violating vehicle and a non-violating vehicle based on the at least one of the trajectory and the color of the signal lamp being red.

3. The method of claim 2, wherein the determining the color of the signal includes:

defining at least two lamp ROIs in each frame, each lamp ROI including a location of a signal lamp displayed on a traffic light on the image plane;

calculating a measure of intensity for pixels located at the each signal lamp within the lamp ROIs;

determining a lamp ROI corresponding to a highest measured intensity;

identifying the signal lamp corresponding to the determined lamp ROI; and

, classifying the displayed traffic light color based on the identified signal lamp.

4. The method of claim 1, wherein the searching the first ROI for the candidate violating vehicle includes:

performing foreground detection within the ROI via one of motion estimation and background estimation with subsequent subtraction.

5. The method of claim 1, wherein the tracking includes:

extracting a set of feature descriptors from pixels corresponding to a region including the detected candidate violating vehicle;

determining if the detected candidate violating vehicle corresponds with one of a tracked vehicle in a previous frame and a new vehicle of interest in a current frame using the extracted set of feature descriptors;

in response to the detected candidate violating vehicle corresponding with the new vehicle of interest, adding the extracted set of feature descriptors to a pool of current feature descriptors; and

,

tracking the pool of current feature descriptors across the sequence of frames.

6. The method of claim 5, wherein the feature descriptors are selected from a group consisting of:

color attributes of the region including the detected candidate violating vehicle;

geometric attributes of the detected candidate violating vehicle;

shape attributes extracted from the region including the detected candidate violating vehicle; and

,

a combination of the above.

7. The method of claim 5, wherein the determining if the detected candidate violating vehicle corresponds with the tracked vehicle in the previous frame includes:

computing a number of currently tracked features in the region including the detected candidate violating vehicle;

comparing the number of tracked features to a predetermined threshold;

in response to the number of tracked features not meeting and exceeding the predetermined threshold, classifying the detected candidate violating vehicle as belonging to the new vehicle of interest;

in response to the number of tracked features meeting and exceeding the predetermined threshold, classifying the detected candidate violating vehicle as belonging to the tracked vehicle in the previous frame.

8. The method of claim 1, wherein the sequence of frames is acquired from a group consisting of:

video captured in relation to when the candidate violating vehicle triggers a sensor in a road near the first ROI;

video captured in relation to when a red signal light is detected;

video captured in relation to when a signal is transmitted to a signal light for turning the signal light red; and

,

a combination of the above.

9. The method of claim 1, wherein the classifying the vehicle includes:

defining a second ROI located after the stop signal on the image plane;

in response to the at least one trajectory including a component in the first and second ROIs, classifying the candidate violating vehicle as belonging to the violating vehicle.

10. The method of claim 1, wherein the acquiring the sequence of frames includes:

triggering the at least one video camera to capture each sequence of frames only when a signal lamp displayed on the stop signal is red.

11. A system for detecting a vehicle running a stop signal positioned at an intersection, the system comprising a traffic regulation enforcement device including a memory and a processor in communication with the processor configured to:

acquire a sequence of frames from at least one video camera monitoring an intersection being signaled by the stop;

define a first region of interest (ROI) including a road region located before the intersection on the image plane;

search the first ROI for a candidate violating vehicle;

in response to detecting the candidate violating vehicle, track at least one trajectory of the detected candidate violating vehicle;

classify the candidate violating vehicle as belonging to one of a violating vehicle and a non-violating vehicle based on the at least one trajectory.

12. The system of claim 11, wherein the processor is further configured to:

in response to the stop signal being a traffic light, determine a color of a signal lamp displayed on the traffic light as belonging to one of red, yellow, and green; and

,

classify the candidate violating vehicle as belonging to one of a violating vehicle and a non-violating vehicle based on the at least one of the trajectory and the color of the signal lamp being red.

13. The system of claim 12, wherein the processor is further configured to:

define at least two lamp ROIs in each frame, each lamp ROI including a location of a signal lamp displayed on a traffic light on the image plane;

calculate a measure of intensity for pixels located at the each signal lamp within the lamp ROIs;

determine a lamp ROI corresponding to a highest measured intensity;

identify the signal lamp corresponding to the determined lamp ROI; and

,

classify the displayed traffic light color based on the identified signal lamp.

14. The system of claim 11, wherein the processor is further configured to:

perform foreground detection within the ROI via one of motion estimation and background estimation with subsequent subtraction.

15. The system of claim 11, wherein the processor is further configured to:

extract a set of feature descriptors from pixels corresponding to a region including the detected candidate violating vehicle;

determine if the detected candidate violating vehicle corresponds with one of a tracked vehicle in a previous frame and a new vehicle of interest in a current frame using the extracted set of feature descriptors;

in response to the detected candidate violating vehicle corresponding with the new vehicle of interest, add the extracted set of feature descriptors to a pool of current feature descriptors; and

,

track the pool of current feature descriptors across the sequence of frames.

16. The system of claim 15, wherein the feature descriptors is selected from a group consisting:

color attributes of the region including the detected candidate violating vehicle;

geometric attributes of the detected candidate violating vehicle;

shape attributes extracted from the region including the detected candidate violating vehicle; and

,

a combination of the above.

17. The system of claim 15, wherein the processor is further configured to:

compute a number of currently tracked features in the region including the detected candidate violating vehicle;

compare the number of tracked features to a predetermined threshold;

in response to the number of tracked features not meeting and exceeding the predetermined threshold, classify the detected candidate violating vehicle as belonging to the new vehicle of interest;

in response to the number of tracked features meeting and exceeding the predetermined threshold, classify the detected candidate violating vehicle as belonging to the tracked vehicle in the previous frame.

18. The system of claim 11, wherein the processor is further configured to acquire the sequence of frames from a group consisting of:

video captured in relation to when the candidate violating vehicle triggers a sensor in a road near the first ROI;

video captured in relation to when a red signal light is detected;

video captured in relation to when a signal is transmitted to a signal light for turning the signal light red; and

,

a combination of the above.

19. The system of claim 11, wherein the processor is further configured to:

define a second ROI located after the stop signal on the image plane;

in response to the at least one trajectory including a component in both the first and second ROIs, classify the candidate violating vehicle as belonging to the violating vehicle.

20. The system of claim 11, wherein the processor is further configured to:

trigger the at least one video camera to capture each sequence of frames only when a signal lamp displayed on the stop signal is red.