Sensor suite and signal processing for border surveillance

• US 20110001657A1
• Filed: 06/08/2007
• Published: 01/06/2011
• Est. Priority Date: 06/08/2006
• Status: Active Grant

• Current Assignee: Aerostar International LLC (CNH Industrial N.V.)
• Original Assignee: Vista Research, Inc. (Guidepoint Global LLC)
• Inventors: Fox, Philip A., Maresca, Joseph W. JR.
• Granted Patent: US 8,330,647 B2
• Time in Patent Office: Days
• US Class Current: 342/107
• CPC Class Codes: G01S 13/522 using transmissions of interrupted pulse modulated waves
• CPC Class Codes: G01S 13/723 by using numerical data
• CPC Class Codes: G01S 13/86 Combinations of radar systems with non-radar systems, e.g. sonar, direction finder
• CPC Class Codes: G01S 13/867 Combination of radar systems with cameras
• CPC Class Codes: G01S 13/886 for alarm systems alarms with electrical actuation G08B13/22
• CPC Class Codes: G01S 7/2922 by using a controlled threshold

Abstract

A land-based Smart-Sensor System and several system architectures for detection, tracking, and classification of people and vehicles automatically and in real time for border, property, and facility security surveillance is described. The preferred embodiment of the proposed Smart-Sensor System is comprised of (1) a low-cost, non-coherent radar, whose function is to detect and track people, singly or in groups, and various means of transportation, which may include vehicles, animals, or aircraft, singly or in groups, and cue (2) an optical sensor such as a long-wave infrared (LWIR) sensor, whose function is to classify the identified targets and produce movie clips for operator validation and use, and (3) an IBM CELL supercomputer to process the collected data in real-time. The Smart Sensor System can be implemented in a tower-based or a mobile-based, or combination system architecture. The radar can also be operated as a stand-alone system.

Claims

1. A method for detection of the presence of at least one target on land using a non-coherent radar, comprising the steps of:

(a) transmitting a plurality of non-coherent radar pulses toward an area of interest;

(b) receiving the radar return scattered from the area of interest;

(c) processing said return in at least one range-azimuth resolution cell in the area of interest to differentiate the return from the target from the return from system and background noise and clutter; and

(d) applying at least one threshold to differentiate the target from the noise and clutter.

2. The method of claim 1, where the target has changed resolution cells over a period of time.

3. The method of claim 1, where the method includes processing of the radar in a plurality of resolution cells in which the return can be associated with the target.

4. The method of claim 3, where the method includes processing of the radar in a plurality of resolution cells in close proximity to each other.

5. The method of claim 4, where the method includes processing of the radar in a plurality of neighboring resolution cells.

6. The method of claim 1, where the target has a small cross-section.

7. The method of claim 6, where the target is not easily distinguished from the system and background noise/cutter.

8. The method of claim 1, where the radar transmits pulses over 360 degrees.

9. The method of claim 8, where the radar pulses are transmitted in a plurality of separate, adjacent and independent sectors.

10. The method of claim 9, where the sectors overlap.

11. The method of claim 1, where the target of interest is not moving.

12. The method of claim 11, where the target is detected by comparing the radar return obtained with the target present in the resolution cell to another measurement made in that resolution cell without the target present.

13. The method of claim 1, where a statistical distribution of the radar return is used to set a threshold with a known probability of detection and a known probability of false alarm.

14. The method of claim 1, where an empirically-based method of detection is used to distinguish false contacts from actual contacts, using empirical information acquired over a period of time to differentiate when a target is present and when a target is not present.

15. The method of claim 1, using a track-before-detect processing method to identify a moving target.

16. The method of claim 15, where the track-before-detect processing method comprises the steps of:

(a) accumulating the contacts in each resolution cell in the area of interest by setting a threshold for each radar measurement across the area of interest to identify a large but approximately constant number of potential targets; and

(b) determining the presence of the targets from the expected characteristic behavior of actual contacts versus false contacts in one or more resolution cells in close proximity to each other.

17. The method of claim 13, where the detection processing approach comprises steps of:

(a) setting a threshold to achieve a certain probability of detection for each target of interest and a constant probability of false alarm for each detection; and

(b) determining the presence of a target of interest from a decision criterion comprised of an accumulation of one or more threshold excedances in one or more resolution cells in close proximity to each other that is consistent with the expected strength and movement of an actual target.

18. The method of claim 1, where at least one target is selected from the group consisting of:

people, vehicles, and animals.

19. The method of claim 1, where the radar is a X-Band radar.

20. The method of claim 1, where the radar is a Ku-Band radar.

21. The method of claim 1, where the radar is a C-Band radar.

22. The method of claim 1, where the detection is performed using two or more radars located at different positions.

23. The method of claim 1, where a target is tracked over time.

24. The method of claim 1, where a target is moving in an environment in which the target may not always be visible to a radar.

25. The method of claim 1, where a detection of a moving target is declared based on the development of a sensible track.

26. The method of claim 25, where the track is determined from a probability statistic.

27. The method of claim 25 where the speed of the track is consistent with that of a target of interest.

28. The method of claim 25, where the trajectory of the track is consistent with that of a target of interest.

29. The method of claim 1, where the non-coherent radar is also used for tracking or classification of a target.

30. The method of claim 29, where classification is based on a target characteristic that includes a characteristic speed, movement, trajectory, and strength of the target.

31. The method of claim 30, where a statistical distribution of the target characteristics is used for classification.

32. The method of claim 1, where the detection of a moving target is accomplished by determining whether a track derived from the radar return for at least two time intervals is sensible for a target of interest.

33. The method of claim 1, where the radar collects at least three samples at each resolution cell over a period of time before a detection is declared.

34. The method of claim 1, where the radar data is pre-processed to generate a detection field before the return is differentiated.

35. The method of claim 1, where a detection field is generated after conditioning operations are performed on the data, where the conditioning operations include an operation selected from the group consisting of calibration, spatial registration, pre-filtering, constant background removal, and noise background estimation.

36. The method of claim 35, where the spatial registration involves aligning the resolution cells for direct comparison from one scan the next scan.

37. The method of claim 1, where a detection field is generated by segmentation of the data.

38. The method of claim 34, where the pre-processing includes separating the resolution cells into different categories for processing by the type of noise and clutter present.

39. The method of claim 33, where the pre-processing includes separating the resolution cells into those dominated by system/receiver noise from all other types of noise and clutter.

40. The method of claim 32, where the track is determined to be sensible by the level of confidence of a probability statistic applied to the return.

41. The method of claim 32, where target differentiation is accomplished by using a low-amplitude threshold to insure a high probability of detection of targets of interest in the area of interest.

42. The method of claim 41, where a second threshold is used to eliminate the false contacts that are realized from the low threshold and to identify the targets of interest based on the sensibility of the track developed from the systematic presence of the actual contacts amidst the random set of false contacts.

43. The method of claim 40, wherein the probability statistic is defined by a log-likelihood ratio that assumes that the system noise and clutter within each radar cell is substantially constant over a period of time, whereby the target-absent statistics over many scans can be determined and used in the determination of the log-likelihood ratio (LLR) built from the radar output return.

44. The method of claim 40, where the probability statistic is based on Bayesian statistical methods.

45. The method of claim 1, for detection of a land-based target using the output radar return from multiple scans of said radar in the target area collected over a period of time and applying two thresholds, one threshold low enough to identify the presence of one or more actual targets as well as many false contacts, and a second one to detect the presence of said one or more actual targets and to eliminate the false contacts based on an odds ratio applied to the first set of contacts that grows when an actual target is present until it exceeds a second threshold.

46. The method of claim 45, where the odds ratio is computed based on Bayes Rule, which multiplies current track odds ratio by likelihood ratio at each observation to update the odds ratio.

47. A method for surveillance to detect a land target using a non-coherent radar, comprising the steps of claim 1 and further comprising the steps of:

(a) determining the background statistics of the radar measurements in each resolution cell when a target of interest is not present;

(b) using said background statistics in each resolution cell to remove the background in the radar data obtained when the radar return from a target of interst may be present in one or more of the resolution cells;

(c) detecting the presence of a target of interest when said radar return obtained in (b) exceeds some threshold value.

48. The method of claim 1, where the non-coherent radar is used to cue an optical sensor system to look at the target in the field-of-view of said sensor system.

49. The method of claim 48, where the optical sensor system is used to detect, track, or classify the target.

50. The method of claim 48, where the non-coherent radar is used to cue a coherent radar sensor system.

51. The method of claim 50, where the coherent radar sensor system is used to detect, track, or classify the target.

52. The method of claim 50, further including an optical sensor system.

53. The method of claim 48, where the non-coherent radar is used to track or classify the target.

54. The method of claim 49, where classification is based on at least one target characteristic selected from the group consisting of speed, movement, trajectory, and strength of the target.

55. The method of claim 48, where a plurality of said sensor systems located at different positions are used in conjunction with the first measurement system for detection, tracking or classification of the target.

56. The method of claim 48, where an image of the target is sent to a person for validation of the existence and type of target present.

57. The method of claim 48, where the detection of a target moving on land is accomplished using an optical sensor system, further comprising the steps of:

(a) processing the pixels in the field-of-view of the target area of interest to differentiate the return from the target of interest from the return from system and background and environmental noise; and

(b) applying at least one threshold to differentiate the target from said noise.

58. The method of claim 57, where a statistical distribution of the optical return is used to set a threshold with a known probability of detection and a known probability of false alarm.

59. The method of claim 57, where an empirically-based method of detection is used to distinguish false contacts from actual contacts using empirical information acquired over a period of time to differentiate when a target is present and when a target is not present.

60. The method of claim 57, using a track-before-detect processing method.

61. The method of claim 60, where the track-before-detect processing method comprises the steps of:

(a) accumulating the contacts in each pixel in the area of interest by setting a threshold for each optical across the area of interest to identify a large but approximately constant number of potential targets; and

(b) determining the presence of the targets from the expected characteristic behavior of actual contacts versus false contacts in one or more pixels in close proximity to each other.

62. The method of claim 58, where the detection processing approach comprises the steps of:

(a) setting a threshold to achieve a certain probability of detection for each target of interest and a probability of false alarm for each detection; and

(b) determining the presence of a target of interest from a decision criterion comprised of an accumulation of one or more threshold excedances in one or more resolution cells in close proximity to each other that is consistent with the expected strength and movement of an actual target.

63. The method of claim 48, for detection of a land target with an optical sensor system, further comprising the steps of:

(a) determining the background statistics of the optical sensor measurements in each pixel of the image obtained when a target is not present;

(b) using the background statistics in each pixel in the image to remove the background in each pixel obtained when the return from a target may be present in one or more of the pixels; and

(c) detecting the presence of the target when said optical sensor return obtained in (b) exceeds a pre-defined threshold level.