Image based detecting system and method for traffic parameters and computer program product thereof

• US 9,058,744 B2
• Filed: 03/11/2011
• Issued: 06/16/2015
• Est. Priority Date: 12/09/2010
• Status: Active Grant

• Current Assignee: Industrial Technology Research Institute
• Original Assignee: Industrial Technology Research Institute
• Inventors: Huang, Chung-Hsien, Wu, Ruei-Cheng
• Primary Examiner(s): Bali, Vikkram
• Application Number:
  US13/046,689
• Publication Number:
  US 20120148094A1
• Time in Patent Office: 1,558 Days
• Field of Search: 382/103, 382/104, 382/107, 348/135, 348/148, 348/149
• US Class Current: 1/1
• CPC Class Codes: G06V 20/54 of traffic, e.g. cars on the road, trains or boats
• CPC Class Codes: G08G 1/054 photographing overspeeding vehicles

Abstract

An image-based detecting system for traffic parameters first sets a range of a vehicle lane for monitoring control, and sets an entry detection window and an exit detection window in the vehicle lane. When the entry detection window detects an event of a vehicle passing by using the image information captured at the entry detection window, a plurality of feature points are detected in the entry detection window, and will be tracked hereafter. Then, the feature points belonging to the same vehicle are grouped to obtain at least a location tracking result of single vehicle. When the tracked single vehicle moves to the exit detection window, according to the location tracking result and the time correlation through estimating the information captured at the entry detection window and the exit detection window, at least a traffic parameter is estimated.

Claims

1. An image-based detecting system for traffic parameters, comprising a processor which further includes:

a vehicle lane region of interest (ROI) setting module, setting a monitored range on a vehicle lane, and setting an entry detection window and an exit detection window on said vehicle lane;

a vehicle passing event detection module, detecting whether a vehicle passing event has occurred by using image information captured at said entry detection window;

a feature point detection and tracking module, detecting a plurality of feature points detection detected within said entry detection window when a vehicle passing event is being detected, and for tracking said plurality of detected feature points by time;

a feature point grouping module, grouping a plurality of feature points to obtain a location tracking result of a single vehicle; and

a traffic parameter estimation module, when a tracked single vehicle moves to said exit detection window, said traffic parameter estimation module estimating at least a traffic parameter according to said location tracking result of said single vehicle and by estimating temporal correlation of information captured in said entry and said exit detection windows;

wherein said feature point grouping module uses a hierarchical feature point grouping architecture to group feature points belonging to a same vehicle to obtain said location tracking result of said single vehicle, and said hierarchical feature point grouping architecture includes, from bottom to top, a point level, a group level and an object level, said feature point grouping module merges similar feature points and rejects erroneous noise feature points between the point level and the group level, and merges groups with motion consistency and spatial-temporal consistency into a moving object between the group level and the object level.

2. The system as claimed in claim 1, said system further includes an image capturing device for continuously capturing a plurality of lane images.

3. The system as claimed in claim 1, wherein said vehicle passing event detection module obtains an image template at said entry detection window at each time point, extracts a plurality of statistically significant feature points from each image template and uses a support vector machine (SVM) classifier to classify the obtained image templates into three different classes including vehicle, shadow and road.

4. The system as claimed in claim 3, wherein said plurality of statistically significant feature points include at least three gray scale based features and two edge based features.

5. The system as claimed in claim 3, wherein said SVM classifier includes at least a first level SVM classifier and a second level SVM classifier, said first level SVM classifier divides said obtained image templates into road or non-road image templates, and said non-road image templates are further divided by said second level SVM classifier into vehicle and shadow.

6. An image-based detecting method for traffic parameters, applicable to a traffic parameter detecting system, said method comprising:

setting a monitored range on a vehicle lane, and setting an entry detection window and an exit detection window in said vehicle lane;

detecting whether an event of a vehicle passing occurs by using image information captured at said entry detection window, and when said event of a vehicle passing is detected at said entry detection window, detecting a plurality of feature points in said entry detection window and tracking said plurality of feature points being tracked hereafter;

grouping said feature points belonging to a same vehicle to obtain at least a location tracking result of a single vehicle; and

estimating at least a traffic parameter when said single vehicle moves to said exit detection window, according to said location tracking result and temporal correlation through estimating information captured at said entry detection window and said exit detection window;

wherein said method groups feature points belonging to a same vehicle to obtain said location tracking result of said single vehicle by using a hierarchical feature point grouping architecture, and said hierarchical feature point grouping architecture includes, from bottom to top, a point level, a group level and an object level, said method merges similar feature points and rejects erroneous noise feature points between said point level and said group level, and merges groups with motion consistency and spatial-temporal consistency into a moving object between said group level and said object level.

7. The method as claimed in claim 6, wherein operations between levels of said hierarchical feature point grouping architecture include point-level grouping, group-level rejection, and group-level mergence.

8. The method as claimed in claim 7, wherein said point-level grouping further includes:

when a vehicle passing event is detected in a plurality of successive image frames, grouping a plurality of feature points detected by said plurality of successive image frames into a feature point group.

9. The method as claimed in claim 7, wherein said group-level rejection further includes:

tracking each feature point group of said group level through a Kalman filter; and

rejecting a feature point by using prediction and updating capabilities of said Kalman filter and by using a distant equation condition as a basis.

10. The method as claimed in claim 7, wherein said group-level mergence further includes:

taking motion consistency MC(G_p, G_q) and spatial-temporal consistency ST(G_p, G_q) of two neighboring groups G_p, G_q into account; and

a condition for merging groups G_p and G_q into an object being;

w·

MC(G_p,G_q)+(1−

w)·

ST(G_p,G_q)>

γ

wherein w is a weight variable, and y is a threshold set by a user.

11. The method as claimed in claim 10, wherein said group-level mergence further includes:

computing cross correlation of moving distances of trajectories of two groups G_p, G_q during a past duration as MC(G_p, G_q).

12. The method as claimed in claim 10, wherein said group-level mergence further includes:

expressing ST(G_p, G_q) as a foreground rate FR(G_p, G_q), and said foreground rate FR(G_p, G_q) being equal to (foreground area)/(area between G_p and G_q).

13. The method as claimed in claim 6, wherein said method further includes:

combining correlation of signals observed at said entry and said exit detection windows and similarity of average vehicle traffic time of said vehicle lane to estimate average speed on said vehicle lane.