Image processing techniques for a video based traffic monitoring system and methods therefor

• US 20060274917A1
• Filed: 08/15/2006
• Published: 12/07/2006
• Est. Priority Date: 11/03/1999
• Status: Active Grant

• Current Assignee: Cet Technoloiges Pte., Ltd. (Singapore Technologies Engineering Limited)
• Original Assignee: Cet Technoloiges Pte., Ltd. (Singapore Technologies Engineering Limited)
• Inventors: Chong, Chee Chung, Gu, Ming Kun, Ang, Kim Siah, Ng, Yew Liam
• Granted Patent: US 7,460,691 B2
• Time in Patent Office: Days
• US Class Current: 382/103
• CPC Class Codes: G06V 10/255 Detecting or recognising potential candidate objects based on visual cues, e.g. shapes
• CPC Class Codes: G06V 20/54 of traffic, e.g. cars on the road, trains or boats
• CPC Class Codes: G06V 2201/08 Detecting or categorising vehicles
• CPC Class Codes: G08G 1/0104 Measuring and analyzing of parameters relative to traffic conditions
• CPC Class Codes: G08G 1/04 using optical or ultrasonic detectors
• CPC Class Codes: H04N 7/18 Closed-circuit television [CCTV] systems, i.e. systems in which the video signal is not broadcast

Abstract

The present disclosure relates to a number of inventions directed, generally, to the application of image processing techniques to traffic data acquisition using video images. The inventions reside in a traffic monitoring system, the basic function of which is for traffic data acquisition and incident detection. More specifically, the application of image processing techniques for the detection of vehicle, from sequence of video images, as well as the acquisition of traffic data and detection of traffic incident. In one aspect, the present invention provides a method of processing images received from a video based traffic monitoring system. In another aspect, the present invention is directed to a Region Of Interest (ROI) for detection of a moving vehicle and a further aspect is directed to a method of detecting day or night status in a traffic monitoring system. It'"'"'s The application of various algorithms to a video based traffic monitoring system is also considered inventive. Other inventive aspects of the present traffic monitoring system are outlined in the claims.

Claims

1. A region of interest for a traffic monitoring system comprising:

first and second sections being substantially aligned with a respective lane of traffic to be monitored;

a profile-speed-zone for the extraction of vehicle speed if a vehicle is detected; and

a vehicle-detection window for the detection of the presence of the vehicle within the window, the vehicle-detection window partially overlapping the profile-speed zone.

2. A method of reducing the visual impact of a shadow in the processing of images in a video based traffic monitoring system, comprising:

determining at least one area of the image having a relatively constant intensity value;

designating the area as a portion of a shadow area; and

removing visual impact of the area from the image.

3. A method as claimed in claim 2, in which, determining step, the at least one area has a substantially homogenous texture.

4. A method as claimed in claim 2, wherein the visual impact is removed by substantially equalizing the contrast of the area with that of another region proximate the area.

5. A method as claimed in claim 2, further including the step of further reducing the visual impact of the areas from the image.

6. A method of detecting a vehicle within a chevron area in a traffic monitoring system, the method comprising the steps of:

applying a gray level co-occurrence matrix for the extraction of texture features of a region of interest; and

calculating two texture measurements using the gray level co-occurrence matrix, the two texture measurements comprising an angular second moment and a contrast.

7. A method as claimed in claim 6, further including the step of generating the gray level co-occurrence matrix from all of the pixels within the region of interest.

8. A method as claimed in claim 6, further including the steps of:

comparing the angular second moment ASM and the contrast CON, with a background angular second moment ASM_B and a background contrast CON_B, the background angular second moment and background contrast measured when no vehicle is present,); and

determining if (|ASM_B−

ASM|>

ASM_TH−

AND|CON_B−

CON|<

+(CON_Th),

wherein if (|ASM_B−

ASM|>

ASM_TH−

AND|CON_B−

CON|<

+(CON_Th), then a vehicle is not present.

9. A method as claimed in claim 8, further including the step of updating background features if a vehicle is not present.

10. A method of calculating vehicle speed in a traffic monitoring system, the method including the step of using convolution between two edge-profiles for the extraction of the vehicle speed.

11. A method as claimed in claim 10, in which the two edge-profiles are obtained from consecutive frames from the system.

12. A method as claimed in claim 10, further including the step of:

obtaining edge values E(x,y) for all pixels (x,y), within the profile speed zone according to;

$E<br><br>\left(x,y\right)=\left(1-K\right)*\mathrm{EH}<br><br>\left(x,y\right)+K*\mathrm{EV}<br><br>\left(x,y\right)$$\mathrm{wherein}$$\mathrm{EH}<br><br>\left(x,y\right)=<br><br>\underset{m=-1,\text{\hspace{1em}<br><br>}<br><br>n=-1}{\overset{m=1,\text{\hspace{1em}<br><br>}<br><br>n=1}{\sum <br><br>}}<br><br>\mathrm{SH}<br><br>\left(m+1,n+1\right)*1<br><br>\left(x+m,y+n\right)<br><br>$$\mathrm{and}$$\mathrm{EV}<br><br>\left(x,y\right)=<br><br>\underset{m=-1,\text{\hspace{1em}<br><br>}<br><br>n=-1}{\overset{m=1,\text{\hspace{1em}<br><br>}<br><br>n=1}{\sum <br><br>}}<br><br>\mathrm{SV}<br><br>\left(m+1,n+1\right)*1<br><br>\left(x+m,y+n\right)<br><br>$

and wherein SH and SV are 3×

3 matrices for the extraction of the horizontal and vertical edge, respectively.

13. A method as claimed in claim 10, further including the step of:

generating an edge-profile in accordance with

$E_{\mathrm{AVE}}<br><br>\left(y\right)|\mathrm{frame}=f)=\frac{\underset{x=1}{\overset{N<br><br>\left(y\right)}{\sum <br><br>}}<br><br>E<br><br>\left(x,y\right)}{N<br><br>\left(y\right)}$

wherein E_AVE(y) is the average edge-density of each row of pixels.

14. A method as claimed in claim 13, further including the step of:

determining speed if the state of the vehicle detection window is Activate.

15. A method as claimed in claim 14, wherein the convolution used is by performing convolution for functions E_AVE(y|frame=f) and E_AVE(y|frame=f−

1) and

C(z)=Σ

E_AVE(y|frame=f)*E_AVE(y−

z|frame=f−

1)

for all y.

16. A method as claimed in claim 15 further including the step of, for all z, finding a maximum of C(z), C_max(z), wherein:

vehicle speed α

Z_max\C(Z_max=C_max(Z)

17. A method as claimed in claim 16 further including the step of updating E_AVE(y|frame=f−

1);

for all y;

E_AVE(y|frame=f−

1)=E_AVE(y|frame=f).

18. A method as claimed in claim 17, wherein an intensity profile of the region of interest is used in place of the edge-profile.

19. A method of determining vehicle direction in a traffic monitoring system as claimed in claim 18, further comprising the step of using the polarity sign of an offset obtained from the convolution method.

20. A method as claimed in claim in 19, wherein a positive polarity indicates that the vehicle is travelling in a direction of traffic flow, and wherein a negative polarity indicates that the vehicle is travelling in a direction opposite to the direction of traffic flow.