Bicycle presence detection in a roadway using video data analytics

• US 8,933,818 B2
• Filed: 03/13/2013
• Issued: 01/13/2015
• Est. Priority Date: 03/13/2013
• Status: Active Grant

• Current Assignee: Iteris Incorporated
• Original Assignee: Iteris Incorporated
• Inventors: Whiting, Michael T., Gao, Yan, Hwang, Robert J., Lam, Wing, Kreter, Todd W., Linton, Matthew
• Primary Examiner(s): HOFSASS, JEFFERY A
• Application Number:
  US13/798,798
• Publication Number:
  US 20140266806A1
• Time in Patent Office: 671 Days
• Field of Search: 340/933, 340/937, 701/300, 382/103, 382/104
• US Class Current: 340/937
• CPC Class Codes: G08G 1/015 with provision for distinguishing between two or more types of vehicles, e.g. between motor-cars and cycles
• CPC Class Codes: G08G 1/04 using optical or ultrasonic detectors

Abstract

A vehicular observation and detection apparatus and system incorporates a detection framework using video analysis to differentiate between motorized vehicles and bicycles for improved traffic flow and safety at intersections. The detection framework creates virtual zones overlaid on lanes of a roadway and analyzes input data representing objects in the virtual zones collected from one or more cameras positioned at or near the roadway.

Claims

1. A method of detecting a presence of bicycles in a roadway for traffic signal controller management, comprising:

creating a virtual bicycle zone in one or more lanes of a roadway near a traffic intersection;

generating, for each virtual bicycle zone, a bicycle detection strip at a specified portion adjacent to each virtual bicycle zone, the bicycle detection strip having a substantially similar shape as the bicycle zone;

detecting motion inside the bicycle detection strip using an object detection means; and

initiating one or more data processing functions embodied in a plurality of instructions configured to classify a type of object present as the object travels through the bicycle detection strip, the one or more data processing functions including

determining whether the object present in the bicycle detection strip is either a wide object or a narrow object by analyzing a plurality of pixels that experience changes from frame to frame in a video data stream as the object passes through the bicycle detection strip, wherein a narrow object is initially classified as a bicycle, and a wide object is initially classified as a motorized vehicle; and

analyzing one or more additional characteristics of the object inside the virtual bicycle zone to confirm or reject the initial classification of the object, the one or more additional characteristics including gray scale changes in images from the video data stream, comparison of predicted behavior of the object, and a plurality of detection rules based on additional data from the object detection means differentiating bicycles from motorized vehicles.

2. The method of claim 1, further comprising expanding the virtual bicycle zone to account for a camera angle relative to the one or more lanes of a roadway.

3. The method of claim 1, wherein each bicycle detection strip has a unique height and unique width relative to the virtual bicycle zone.

4. The method of claim 1, wherein the bicycle detection strip is automatically generated every time a virtual bicycle zone is created.

5. The method of claim 1, wherein a size of the bicycle detection strip varies depending on a location of the virtual bicycle zone in each lane, a size of the virtual bicycle zone, and a number and location of surrounding detection zones in the one or more lanes of the roadway.

6. The method of claim 1, wherein the object detection means is at least one video camera positioned above or near the one or more lanes of a roadway.

7. The method of claim 1, wherein the additional data from the object detection means differentiating bicycles from motorized vehicles is a speed of the object.

8. The method of claim 1, further comprising creating a virtual vehicle zone for detecting a presence of motorized vehicles in the one or more lanes of the roadway.

9. A vehicular detection apparatus comprising:

a virtual bicycle zone overlaid on one or more lanes of a roadway near a traffic intersection;

a bicycle detection strip applied at a specified portion adjacent to each virtual bicycle zone, the bicycle detection strip having a substantially similar shape as the virtual bicycle zone;

a plurality of detection devices arranged relative to the one or more lanes of the roadway surface to sense a presence of an object in the bicycle detection strip and generate one or more signals for determination of whether the object is a bicycle or a motorized vehicle;

a plurality of modules configured to apply data processing functions to the one or more signals generated by the plurality of detection devices, the one or more data processing functions embodied in a plurality of instructions configured to classify a type of the object present as the object travels through the bicycle detection strip, the data processing functions configured to

determine whether the object present in the bicycle detection strip is either a wide object or a narrow object by analyzing a plurality of pixels that experience changes from frame to frame in a video data stream as the object passes through the bicycle detection strip, wherein a narrow object is initially classified as a bicycle, and a wide object is initially classified as a motorized vehicle; and

analyze one or more additional characteristics of the object inside the virtual bicycle zone to confirm or reject the initial classification of the object, the one or more additional characteristics including gray scale changes in images from the video data stream, comparison of predicted behavior of the object, and a plurality of detection rules based on additional data from the object detection means differentiating bicycles from motorized vehicles.

10. The apparatus of claim 9, wherein the plurality of detection devices include at least one video camera positioned at an angle relative the one or more lanes in the roadway.

11. The apparatus of claim 9, wherein the virtual bicycle zone and the bicycle detection strip are generated by a detection processor located proximate to an intersection traffic controller responsible for traffic signal management at the intersection.

12. The apparatus of claim 9, further comprising an expansion of the virtual bicycle zone to account for a camera angle relative to the one or more lanes of the roadway.

13. The apparatus of claim 9, wherein the bicycle detection strip has a unique height and a unique width relative to the virtual bicycle zone.

14. The apparatus of claim 9, wherein the bicycle detection strip is automatically generated every time a virtual bicycle zone is created.

15. The apparatus of claim 9, further comprising a virtual vehicle zone for detecting a presence of motorized vehicles, overlaid on the one or more lanes of the roadway.

16. The apparatus of claim 15, wherein the virtual bicycle zone and the virtual vehicle zone overlap in at least one lane of the roadway.

17. A method of performing traffic management in an intended area, comprising:

collecting input data triggered by movement of an object into a bicycle detection strip adjacent to a virtually-created bicycle zone in each lane of a roadway proximate to a traffic intersection, the input data collected from a plurality of detection devices located in or near each lane of the roadway;

performing one or more data processing functions embodied in a plurality of instructions configured to classify a type of object from the input data as the object travels through the bicycle detection strip, the one or more data processing functions including

determining whether the object present in the bicycle detection strip is either a wide object or a narrow object by analyzing a plurality of pixels that experience changes from frame to frame in a video data stream as the object passes through the bicycle detection strip, wherein a narrow object is initially classified as a bicycle, and a wide object is initially classified as a motorized vehicle; and

analyzing one or more additional characteristics of the object inside the virtually-created bicycle zone to confirm or reject the initial classification of the object, the one or more additional characteristics including gray scale changes in images from the video data stream, comparison of predicted behavior of the object, and a plurality of detection rules based on additional data from the object detection means differentiating bicycles from motorized vehicles,

generating output data in one more signals to a traffic signal controller to adjust a timing of at least one traffic light in the traffic intersection where a bicycle is detected.

18. The method of claim 17, further comprising generating the virtually-created bicycle zone and the bicycle detection strip for each lane of the roadway.

19. The method of claim 18, further comprising automatically generating the bicycle detections strip each time the virtually-created bicycle zone is created at a specified portion adjacent to each virtually-created bicycle zone, the bicycle detection strip having a substantially similar shape as the virtually-created bicycle zone, wherein the bicycle detection strip has a unique height and unique width relative to the virtually-created bicycle zone.

20. The method of claim 17, further comprising expanding the virtually-created bicycle zone to account for a camera angle relative to the one or more lanes of the roadway.

21. The method of claim 17, further comprising detecting the object in the bicycle detection strip using the plurality of detection devices, the plurality of detection devices including at least one video camera.

22. The method of claim 17, further comprising generating a virtually-created vehicle zone for detecting a presence of motorized vehicles.

23. The method of claim 22, wherein the virtually-created bicycle zone and the virtually-created vehicle zone overlap in at least one lane of the roadway.