Video surveillance system

• US 7,859,564 B2
• Filed: 08/07/2007
• Issued: 12/28/2010
• Est. Priority Date: 06/04/2003
• Status: Active Grant

• Current Assignee: Model Software Corporation
• Original Assignee: Model Software Corporation
• Inventors: Kelly, John J III, Miller, J. Matthew III
• Primary Examiner(s): CZEKAJ, DAVID J
• Application Number:
  US11/890,515
• Publication Number:
  US 20080211907A1
• Time in Patent Office: 1,239 Days
• Field of Search: 348/143, 348150-155, 348169-172
• US Class Current: 348/143
• CPC Class Codes: G06T 2207/30196 Human being; Person
• CPC Class Codes: G06T 2207/30241 Trajectory
• CPC Class Codes: G06T 7/277 involving stochastic approaches, e.g. using Kalman filters
• CPC Class Codes: G06V 10/28 Quantising the image, e.g. histogram thresholding for discrimination between background and foreground patterns
• CPC Class Codes: G06V 20/52 Surveillance or monitoring of activities, e.g. for recognising suspicious objects recognising microscopic objects G06V20/69
• CPC Class Codes: G08B 13/19613 Recognition of a predetermined image pattern or behaviour pattern indicating theft or intrusion
• CPC Class Codes: G08B 13/19645 Multiple cameras, each having view on one of a plurality of scenes, e.g. multiple cameras for multi-room surveillance or for tracking an object by view hand-over
• CPC Class Codes: H04N 7/181 for receiving images from a plurality of remote sources

Abstract

Viewing a three dimensional area from numerous camera angles at different exposures using intensity and chromaticity data at the different exposures to create a pixel model for each pixel. A current image is compared with the background model to find pixels that have changed from their pixel model. These novel pixels are processed using contiguous region detection and grouped into foreground figures. For each camera, software extracts features from each foreground figure. A central processor maintains an object model for each foreground figure. A graphical user interface displays the relative location of foreground figures in a world model. The location and identification of the foreground figures is checked against a table of permissions so as to selectively generate an alert. If a figure leaves or is about the leave a cell, the invention accounts for its approximate position and notifies adjacent cells of the expected arrival of the foreground figure.

Claims

1. A method for video surveillance, the method comprising the following steps:

analyzing, in a computer, video data from a first perspective to create a first background model;

analyzing, in a computer, video data from a second perspective to create a second background model comparing video data from the first perspective with the first background model so as to detect a first set of novel pixels;

comparing, in a computer, video data from the second perspective with the second background model so as to detect a second set of novel pixels;

comparing, in a computer, features from the first set of novel pixels with object models in a world model;

comparing, in a computer, features from the second set of novel pixels with models in the world model;

identifying from features from the first set of novel pixels and features from the second set of novel pixels an object in a world model;

estimating a position, velocity and acceleration of an object based on said features from the first set of novel pixels; and

estimating a new position of the identified object based on a volume of space where that object could possibly be at a current time, said volume being based on an intersection of a previous position estimate of the object and said estimates of velocity and acceleration associated with the position estimate based on said first set of novel pixels with a position estimate that is based on said second set of novel pixels.

2. A method according to claim 1, wherein the world model is updated before comparing the features from the second set of novel pixels with object models in the world model.

3. A method according to claim 1, further comprising placing features from the first set of novel pixels into queue, and placing features from a second set of novel pixels into the queue, wherein the queue is a time priority queue.

4. A method according to claim 3, wherein a feature from the first set of novel pixels is not processed until the features from the second set of novel pixels is available.

5. A method according to claim 4, the method further comprising the step of determining whether to update one of the object models or add an object model to the world model.

6. A method according to claim 1, wherein a feature from the second set of novel pixels is processed, said second set of novel pixels being obtained from second images taken after first images, and the world model is reverted to the first background model before the second set of novel pixels is processed.

7. A method according to claim 6, the method further comprising the step of determining whether to update one of the object models or add an object model to the world model.