SYSTEM AND METHOD FOR ON-ROAD TRAFFIC DENSITY ANALYTICS USING VIDEO STREAM MINING AND STATISTICAL TECHNIQUES

• US 20130073192A1
• Filed: 09/13/2012
• Published: 03/21/2013
• Est. Priority Date: 09/20/2011
• Status: Active Grant

• Current Assignee: Infosys Limited
• Original Assignee: Infosys Limited
• Inventors: Pisipati, Radha Krishna, Hota, Rudra Narayan, Jonna, Kishore
• Granted Patent: US 8,942,913 B2
• Time in Patent Office: Days
• US Class Current: 701/118
• CPC Class Codes: G08G 1/04 using optical or ultrasonic detectors

Abstract

A method and system for analyzing on-road traffic density are provided. The method involves allowing a user to select a video image capturing device and coordinates in a video image frame captured by the video image capturing device such that the coordinates form a region of interest (ROI). The ROI is processed to generate a confidence value and a traffic density value. The traffic density value is compared with a first set of threshold values. Based on the comparison, the traffic density values at different instants in a time window are displayed to enable monitoring of the traffic trend.

Claims

1. A computer-implemented method for analyzing on-road traffic density comprising:

a. allowing a user to select a video image capturing device from among a plurality of video image capturing devices;

b. allowing the user to select coordinates in one of the one or more video image frames of an on-road traffic scenario captured by the selected video image capturing device such that the coordinates form a closed region of interest (ROI);

c. processing the ROI, wherein the processing comprises;

i. segmenting the ROI into one or more overlapping sub-windows,

ii. converting the one or more overlapping sub-windows into one or more feature vectors through a textural feature extraction technique;

d. generating a traffic confidence value or no traffic confidence value for each of the feature vectors by a traffic density classifier;

e. computing a traffic density value depending on a number of sub-windows with high traffic and total number of sub-windows within the ROI;

f. comparing the traffic density value with a first set of threshold values to categorize the video image frame as having low, medium or high traffic; and

g. displaying traffic density values at different instants in a time window to enable monitoring of a traffic trend.

2. The method according to claim 1, further comprising, based on the estimated traffic density value:

a. estimating a traffic state at a junction;

b. estimating a travel time between any two consecutive junctions on a route, wherein the route includes a plurality of junctions;

c. planning an optimized route between a selected source and a selected destination on the route; and

d. analyzing an impact of congestion at one junction on another junction on the route.

3. The method according to claim 2, wherein the step of estimating the traffic state at a junction comprises:

a. receiving traffic density values of the video image frames captured by the selected video image capturing device for a time window from a database; and

b. comparing the traffic density values with a second set of threshold values to classify traffic state of the time window into one of a plurality of predefined traffic states, wherein the second set of threshold values include a minimum threshold value and a maximum threshold value.

4. The method according to claim 3, wherein the plurality of predefined traffic states comprise a free state, a fluid state and a congestion state.

5. The method according to claim 4, wherein the traffic state of the time window is classified as:

a. free state if the traffic density values in the time window is below the minimum threshold value of the second set of threshold values;

b. fluid state if the traffic density values in the time window are between the maximum and minimum threshold values of the second set of threshold values, and

c. congestion state if the traffic density values in the time window are above the maximum threshold value of the second set of threshold values.

6. The method according to claim 2, wherein the step of estimating the travel time is performed by adding time taken to travel between the any two consecutive junctions on the route and the traffic states between the any two consecutive junctions on the route at different instants of time.

7. The method according to claim 2, wherein the step of planning the optimized route between the selected source and the selected destination comprises finding an optimum path between the selected source and the selected destination using one of static estimation and dynamic estimation.

8. The method according to claim 7, wherein the static estimation identifies a best route based on least time taken to reach the destination and the traffic density values of the junctions between the selected source and the selected destination.

9. The method according to claim 7, wherein the dynamic estimation identifies the best route by utilizing one of graph theory algorithms, such as Kruskal'"'"'s algorithm and Dijkstra'"'"'s algorithm.

10. The method according to claim 2, wherein the step of analyzing the impact of congestion comprises:

a. choosing a congestion time window tc;

b. computing travel time t1 between a pair of junctions J1 and J2 using historical data;

c. obtaining traffic density values Dl for the junction J1 between timestamps t and t+tc, and traffic density values D2 for the junction J2 between timestamps t+t1 and t+t1+tc, where t is the time at any given instant;

d. finding a correlation value between the traffic density values D1 and D2; and

e. comparing the correlation value with a third set of threshold values to categorize the impact of congestion as one of high, medium, low and negative.

11. The method according to claim 10, wherein the third set of threshold values comprises a minimum threshold value below which the congestion impact at J2 on J1 is low and a maximum threshold value above which the congestion impact at J2 on J1 is high.

12. The method according to claim 10, wherein the correlation value is negative when congestion impact is present at J1 due to traffic at J2.

13. The method according to claim 1, wherein the step of selecting the ROI is preceded by allowing the user to select one among a plurality of field of views of the selected video image capturing device.

14. The method according to claim 1, wherein the ROI is a flexible convex shaped polygon.

15. The method according to claim 1, wherein the step of processing the ROI comprises:

a. enhancing contrast in a shadowed region in the ROI; and

b. smoothing the ROI for image noise reduction prior to the step of segmentation of the ROI into sub-windows.

16. The method according to claim 1, wherein the textural feature extraction technique utilizes a histogram of Oriented Gradient descriptor in the sub-windows for converting the sub-windows into feature vectors.

17. The method according to claim 1, wherein the step of generating the positive classification confidence and the negative classification confidence comprises utilizing a non-linear interpolation to provide weightage to the sub-windows based on the distance of the sub-windows from the field of view.

18. The method according to claim 1, wherein the traffic density classifier is pre-trained with a number of manually selected video image data with and without the presence of traffic objects.

19. The method according to claim 1, wherein the first set of threshold values comprise a minimum threshold value below which the traffic density is low and a maximum threshold value above which the traffic density is high.

20. The method according to claim 1, further comprising generating an alarm message when the traffic density value exceeds the first set of threshold values.

21. A method for re-training a traffic density classifier comprising:

a. collecting a set of misclassified video image frames captured by an image capturing device from among a plurality of image capturing devices; and

b. utilizing a reinforcement learning to train the traffic density classifier with a valid set of video image frames corresponding to predefined settings of the image capturing device.

22. The method according to claim 21, wherein the step of collecting a set of misclassified video image frames comprises cross-validating the classified video image frames with a master classifier, where the master classifier is pre-trained with video image frames of multiple texture and color features.

23. The method according to claim 21, wherein the predefined settings of the image capturing device comprise view angle, distance, and height.

24. A system for analyzing on-road traffic density, the system comprising:

a. a plurality of video image capturing devices configured to capture one or more video image frames of an on-road traffic scenario;

b. a user interface device configured to;

i. select a video image capturing device from among the plurality of video image capturing devices;

ii. select coordinates in one of the one or more video image frames captured by the selected video image capturing device such that the coordinates form a closed region of interest (ROI).

c. a processing engine configured to;

i. segment the ROI into one or more overlapping sub-windows; and

ii. convert the one or more overlapping sub-windows into one or more feature vectors through a textural feature extraction technique;

d. a traffic density classification engine configured to;

i. generate a traffic classification confidence value or a no-traffic classification confidence value for each of the feature vectors;

ii. compute a traffic density value based on a number of sub-windows with high traffic and total number of sub-windows within the ROI; and

iii. compare the traffic density value with a first set of threshold values to categorize the video image frame as having low, medium or high traffic; and

e. a display unit configured to display traffic density values at different instants in a time window to enable monitoring a traffic trend.

25. The system according to claim 24, further comprising a traffic density analyzer configured to:

a. estimate a traffic state at a junction;

b. estimate a travel time between any two consecutive junctions on a route, wherein the route includes a plurality of junctions;

c. plan an optimized route between a selected source and a selected destination on the route; and

d. analyze an impact of congestion at one junction on another junction on the route.

26. The system according to claim 25, wherein the traffic density analyzer estimates the traffic state at a junction by:

a. receiving traffic density values of the video image frames captured by the selected video image capturing device for a time window from a database;

b. comparing the traffic density values with a second set of threshold values to classify the traffic state of the time window into one of a plurality of predefined traffic states, wherein the second set of threshold values include a minimum threshold value and a maximum threshold value.

27. The system according to claim 26, wherein the plurality of predefined traffic states comprise a free state, a fluid state and a congestion state.

28. The system according to claim 26, wherein the traffic state of the time window is classified as:

a. free state if the traffic density values in the time window is below the minimum threshold value of the second set of threshold values;

b. fluid state if the traffic density values in the time window are between the maximum and minimum threshold values of the second set of threshold values; and

c. congestion state if the traffic density values in the time window are above the maximum threshold value of the second set of threshold values.

29. The system according to claim 25, wherein the traffic density analyzer further estimates the travel time by adding time taken to travel between the any two consecutive junctions on the route and the traffic states between the any two consecutive junctions on the route at different instants of time.

30. The system according to claim 25, wherein the traffic density analyzer further plans an optimized route between the selected source and the selected destination by finding an optimum path between the selected source and the selected destination using one of static estimation and dynamic estimation.

31. The system according to claim 30, wherein the static estimation identifies a best route based on least time taken to reach the selected destination and the traffic density values of the junctions between the selected source and the selected destination.

32. The system according to claim 30, wherein the dynamic estimation identifies the best route by utilizing one of graph theory algorithms, such as Kruskal'"'"'s algorithm and Dijkstra'"'"'s algorithm.

33. The system according to claim 25, wherein the traffic density analyzer further analyzes the impact of congestion at one junction on another junction by:

a. choosing a congestion time window tc;

b. computing a duration of travel time t1 between a pair of junctions J1 and J2 from historical data;

c. obtaining traffic density values Dl for junction J1 between timestamps t and t+tc, and traffic density values D2 for junction J2 between timestamps t+t1 and t+t1+tc, where t is the time at any given instant

d. finding a correlation value between the traffic density values D1 and D2; and

e. comparing the correlation value with a third set of threshold values to categorize a congestion impact as one of high, medium, low and negative.

34. The system according to claim 33, wherein the third set of threshold values comprises a minimum threshold value below which the congestion impact at J2 on J1 is low and a maximum threshold value above which the congestion impact at J2 on J1 is high.

35. The system according to claim 33, wherein there is congestion impact at J1 due to traffic at J2 when the correlation value is negative.

36. The system according to claim 24, wherein the user interface is further configured to allow the user to select one among the plurality of fields of view for the selected video image capturing device.

37. The system according to claim 24, wherein the ROI is a flexible convex shaped polygon.

38. The system according to claim 24, wherein the processing engine is further configured to process a shadowed region in the ROI for contrast enhancement and smoothing the ROI for image noise reduction prior to segmenting the ROI into sub-windows.

39. The system according to claim 24, wherein the textural feature extraction technique utilizes a histogram of Oriented Gradient descriptor in the sub-windows while converting the sub-windows into feature vectors.

40. The system according to claim 24, wherein the traffic density classification engine generates traffic or no-traffic classification confidences by utilizing non-linear interpolation to provide weightage to the sub-windows based on the distance of the sub-windows from a field of view of the selected video image capturing device.

41. The system according to claim 24, wherein the traffic density classification engine is pre-trained with a number of manually selected video image data with and without the presence of traffic objects.

42. The system according to claim 24, wherein the first set of threshold values comprise a minimum threshold value below which the traffic density is low and a maximum threshold above which the traffic density is high.

43. The system according to claim 24, further comprising an alarm notification unit to generate an alarm message when the traffic density value exceeds the first set of threshold values.

44. A system for re-training a traffic density classification engine comprising:

a. a database to collect a set of misclassified video image data of a video image capturing device from among plurality of video image capturing devices;

b. a reinforcement learning engine to train the traffic density classification engine with a valid set of video image data for corresponding to predefined settings of the video image capturing devices.

45. The system according to claim 44, wherein the set of misclassified video image data is obtained by cross-validating classified video image data with a master classifier, where the master classifier is trained with video image data of multiple textures and color features.

46. The system according to claim 45, wherein the predefined settings of the image capturing device comprise view angle, distance, and height.

47. A computer program product for use with a computer, the computer program product comprising a computer usable medium having a computer readable program code embodied therein for analyzing on-road traffic density, the computer readable program code storing a set of instructions configured for:

a. allowing a user to select a video image capturing device from among a plurality of video image capturing devices;

b. allowing the user to select coordinates in one of the one or more video image frames of an on-road traffic scenario captured by the selected video image capturing device such that the coordinates form a closed region of interest (ROI);

c. processing the ROI, wherein the processing comprises;

i. segmenting the ROI into one or more overlapping sub-windows,

ii. converting the one or more overlapping sub-windows into one or more feature vectors through textural feature extraction technique;

d. generating a traffic confidence value or a no-traffic confidence value for each of the feature vectors by a traffic density classifier;

e. computing a traffic density value depending on a number of sub-windows with high traffic and total number of sub-windows within the ROI;

f. comparing the traffic density value with a first set of threshold values to categorize the video image frame as having low, medium or high traffic; and

g. displaying traffic density values at different instants in a time window to enable monitoring of a traffic trend.

48. The computer program product according to claim 47, wherein the instructions based on the estimated traffic density value are further classified to:

a. estimate a traffic state at a junction;

b. estimate a travel time between any two consecutive junctions on a route, wherein the route includes a plurality of junctions;

c. plan an optimized route between a selected source and a selected destination on the route; and

d. analyze an impact of congestion at one junction on another junction on the route.

49. A computer program product for use with a computer, the computer program product comprising a computer usable medium having a computer readable program code embodied therein for re-training a traffic density classification engine, the computer readable program code storing a set of instructions configured for:

a. collecting a set of misclassified video image frames captured by an image capturing device from among a plurality of image capturing devices; and

b. utilizing a reinforcement learning to train the traffic density classifier with a valid set of video image frames corresponding to predefined settings of the image capturing device.

50. The computer program product according to claim 49, wherein the misclassified video image frames are obtained by cross-validating the classified video image frames with a master classifier, where the master classifier is pre-trained with video image frames of multiple textures and color features.