Statistical modeling and performance characterization of a real-time dual camera surveillance system

• US 7,006,950 B1
• Filed: 06/12/2000
• Issued: 02/28/2006
• Est. Priority Date: 06/12/2000
• Status: Expired due to Term

• Current Assignee: Siemens Corp. (Siemens AG)
• Original Assignee: Siemens Corporate Research Incorporated (Siemens AG)
• Inventors: Ramesh, Visvanathan, Greiffenhagen, Michael, Comaniciu, Dorin
• Primary Examiner(s): Broda, Esq., Samuel
• Application Number:
  US09/592,532
• Time in Patent Office: 2,087 Days
• Field of Search: 703 1- 2, 703/6, 382/107, 382/228
• US Class Current: 703/2
• CPC Class Codes: G06T 2207/30196 Human being; Person
• CPC Class Codes: G06T 2207/30232 Surveillance
• CPC Class Codes: G06T 2207/30241 Trajectory
• CPC Class Codes: G06T 7/277 involving stochastic approaches, e.g. using Kalman filters
• CPC Class Codes: G06T 7/77 using statistical methods
• CPC Class Codes: G08B 13/19604 involving reference image or background adaptation with time to compensate for changing conditions, e.g. reference image update on detection of light level change
• CPC Class Codes: G08B 13/19608 Tracking movement of a target, e.g. by detecting an object predefined as a target, using target direction and or velocity to predict its new position
• CPC Class Codes: G08B 13/19628 of wide angled cameras and camera groups, e.g. omni-directional cameras, fish eye, single units having multiple cameras achieving a wide angle view
• CPC Class Codes: G08B 13/19641 Multiple cameras having overlapping views on a single scene
• CPC Class Codes: G08B 13/19643 wherein the cameras play different roles, e.g. different resolution, different camera type, master-slave camera
• CPC Class Codes: G08B 13/19647 Systems specially adapted for intrusion detection in or around a vehicle
• CPC Class Codes: H04N 23/695 Control of camera direction for changing a field of view, e.g. pan, tilt or based on tracking of objects
• CPC Class Codes: H04N 7/181 for receiving images from a plurality of remote sources

Abstract

The present invention relates to a method for visually detecting and tracking an object through a space. The method chooses modules for a restricting a search function within the space to regions with a high probability of significant change, the search function operating on images supplied by a camera. The method also derives statistical models for errors, including quantifying an indexing step performed by an indexing module, and tuning system parameters. Further the method applies a likelihood model for candidate hypothesis evaluation and object parameters estimation for locating the object.

Claims

1. A method for visually locating and tracking an object through a space, comprising the steps of:

restricting an area of the space to be searched to a plurality of regions with a high probability of significant change, the space being defined in images supplied by a camera;

deriving statistical models for errors, including quantifying an indexing step performed by an indexing module, and tuning system parameters, wherein each statistical model is a candidate hypothesis for object location;

applying a likelihood model for candidate hypothesis evaluation; and

locating the object according to a candidate hypothesis evaluated to satisfy the likelihood model.

2. The method of claim 1, wherein restricting the area of the space to be searched further comprises the steps of:

applying a calibration module for determining a static scene;

applying an illumination-invariant module for tracking image transformation; and

applying the indexing module for selecting regions of interest for hypothesis generation.

3. The method of claim 2, further comprising the steps of:

applying a statistical estimation module for estimating a number of objects and their positions; and

applying a foveal camera control module for estimating a plurality of control parameters of a foveal camera based on location estimates and uncertainties.

4. The method of claim 2, further comprising the step of applying a background adaptation module for detecting and tracking the object in dynamically varying illumination situations.

5. The method of claim 1, wherein an application specific module is provided based on a plurality of prior distributions for imposing restrictions on the candidate hypotheses.

6. The method of claim 5, wherein the plurality of prior distributions comprise:

an object geometry model;

a camera geometry model;

a camera error model; and

an illumination model.

7. The method of claim 1, wherein the camera is an omnicamera.

8. The method of claim 1, wherein the object is tracked using a foveal camera.

9. The method of claim 1, wherein the step of deriving statistical models is applied a plurality of times to achieve a given probability of misdetection and false alarm rate.

10. The method claim 9, further comprising the step of validating a theoretical model for the space monitored for determining correctness and closeness to reality.

11. The method of claim 1, wherein the indexing module selects a plurality of regions with a high probability of significant change, motivated by a plurality of two dimensional image priors induced by a plurality of prior distributions in the space, wherein the space is three dimensional.

12. The method of claim 1, further comprising estimating an uncertainty of the object'"'"'s parameters for predicting a system'"'"'s performance and for automating control of the system.

13. The method of claim 1, employed in an automobile wherein the space monitored comprises one of an interior compartment of the automobile and an exterior of the automobile.

14. The method of claim 1, wherein the candidate hypothesis evaluated that satisfies the likelihood model minimizes a Bayes error.

15. A computer program product comprising computer program code stored on a computer readable storage medium for, for locating and tracking objects through a space, the computer program product comprising:

computer readable program code for causing a computer to restrict an area of the space to be search to a plurality of regions with a high probability of significant change within the space;

computer readable program code for causing a computer to derive statistical models for errors, including quantifying an indexing step, and tuning system parameters, wherein each statistical model is a candidate hypothesis for object location;

computer readable program code for causing a computer to apply a likelihood model for candidate hypothesis evaluation; and

computer readable program code for causing a computer to locate the object according to a candidate hypothesis evaluated to satisfy the likelihood model.