System and method for analyzing video from non-static camera

• US 8,036,426 B2
• Filed: 07/28/2010
• Issued: 10/11/2011
• Est. Priority Date: 01/20/2005
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Pankanti, Sharathchandra U., Hampapur, Arun, Senior, Andrew W.
• Primary Examiner(s): KOZIOL, STEPHEN R
• Application Number:
  US12/845,302
• Publication Number:
  US 20100296704A1
• Time in Patent Office: 440 Days
• Field of Search: 382/103, 382/294
• US Class Current: 382/103
• CPC Class Codes: G06T 3/4038 Image mosaicing, e.g. composing plane images from plane sub-images
• CPC Class Codes: G06T 7/246 using feature-based methods, e.g. the tracking of corners or segments
• 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 13/19663 Surveillance related processing done local to the camera
• CPC Class Codes: H04N 7/185 from a mobile camera, e.g. for remote control

Abstract

A novel system and method of treating the output of moving cameras, in particular ones that enable the application of conventional “static camera” algorithms, e.g., to enable the continuous vigilance of computer surveillance technology to be applied to moving cameras that cover a wide area. According to the invention, a single camera is deployed to cover an area that might require many static cameras and a corresponding number of processing units. A novel system for processing the main video sufficiently enables long-term change detection, particularly the observation that a static object has been moved or has appeared, for instance detecting the parking and departure of vehicles in a parking lot, the arrival of trains in stations, delivery of goods, arrival and dispersal of people, or any other application.

Claims

1. A system for analyzing images acquired from a single non-static camera when viewing an observed area, said system comprising:

a memory;

a processor in communication with the memory, wherein the processor is configured to perform a method comprising;

generating, during a first pass of said non-static camera viewing an observed area under surveillance, a reference model from one or more image frames acquired from said camera;

acquiring current image frames during successive passes of said non-static camera viewing said observed area,

applying, to current image frames feature based registration in conjunction with camera position, to determine a respective output image sequence of images, each image of said output image sequence corresponding to a respective reference image of said reference model; and

,

storing, in a memory storage device, successive output images from said imaging device; and

,

analyzing each of said successive output images for detecting a change in a desired feature of the observed area.

3. The system as claimed in claim 1, wherein said method further comprises:

updating, during each successive pass of said non-static camera, one or more reference images of said reference model.

4. The system as claimed in claim 1, wherein said processor is further configured for monitoring said camera position, said position represented as a distance of a current acquired image frame to a current closest image reference of said reference model.

5. The system as claimed in claim 1, wherein said a respective output image is an image frame having the smallest displacement to a corresponding reference model.

6. The system as claimed in claim 1, wherein said distance measure is one or more of:

a camera angular displacement, the number of pixels of translation to achieve image alignment with a reference model, a number of degrees of rotation to achieve alignment, determinant of transformation matrix, or combinations thereof.

7. The system as claimed in claim 1, wherein said processor is further configured for updating a reference model when a camera achieves predetermined positions during each pass.

8. The system as claimed in claim 7, wherein said reference model of reference images corresponds to a set of images comprising separate, non-overlapping views covering said observed area under surveillance.

9. The system as claimed in claim 1, wherein said camera position is represented by current camera control parameter settings including one or more of:

a pan, tilt, zoom camera, rotation, dolly, and translate control parameters.

10. The system as claimed in claim 1, wherein said applying to frames a feature based registration comprises:

tracking points of features in a current reference image, and estimating the image motion in successive acquired current image frames during each pass.

2. A method for analyzing images acquired from a single non-static camera, said method comprising:

generating, during a first pass of said non-static camera viewing an observed area under surveillance, a reference model from one or more image frames acquired from said camera;

acquiring current image frames during successive passes of said non-static camera viewing said observed area,

applying to current image frames a feature based registration in conjunction with a camera position, to determine a respective output image sequence of images, each image of said output image sequence corresponding to a respective reference image of said reference model;

storing, in a memory storage device, successive output images from said imaging device; and

,

analyzing each of said successive output images for detecting a change in a desired feature of the observed area,

wherein a processor device performs one or more said generating, acquiring, applying and analyzing.

11. The method as claimed in claim 2, further comprising:

updating, during each successive pass of said non-static camera, one or more reference images of said reference scene model.

12. The method as claimed in claim 2, further comprising:

monitoring said camera position, said position represented as a distance of a current acquired image frame to a current closest image reference of said reference model.

13. The method as claimed in claim 2, wherein said a respective output image is an image frame having the smallest displacement to a corresponding reference model.

14. The method as claimed in claim 2, wherein said distance measure is one or more of:

a camera angular displacement, the number of pixels of translation to achieve image alignment with a reference model, a number of degrees of rotation to achieve alignment, determinant of transformation matrix, or combinations thereof.

15. The method as claimed in claim 2, wherein said processor is further configured for updating a reference model when a camera achieves predetermined positions during each pass.

16. The method as claimed in claim 15, wherein said reference model of reference images corresponds to a set of images comprising separate, non-overlapping views covering said observed area under surveillance.

17. The method as claimed in claim 2, wherein said camera position is represented by current camera control parameter settings including one or more of:

a pan, tilt, zoom camera, rotation, dolly, and translate control parameters.

18. The method as claimed in claim 2, wherein said applying to frames a feature based registration comprises:

tracking points of features in a current reference image, and estimating the image motion in successive acquired current image frames during each pass.