Method for Incorporating Facial Recognition Technology in a Multimedia Surveillance System

Abstract

Embodiments provide a surveillance system having at least one camera adapted to produce an IP signal, the at least one camera having an image collection device configured for collecting image data, the at least one camera having at least one facial processor configured to execute with digital format image data at least one facial recognition algorithm, execution of the at least one facial recognition algorithm with the digital format image data detecting faces when present in the digital format image data, execution of the at least onefacial recognition algorithm providing for each detected face at least one set of unique facial image data.

Description

CROSS REFERENCE TO RELATED APPLICATION[0001]This application is a Continuation of application Ser. No. 10 / 719,792, filed on Nov. 21, 2002, titled “Method for Incorporating Facial Recognition Technology in a Multimedia Surveillance System”, the entire contents of which are incorporated herein by reference.BACKGROUND OF THE INVENTION[0002]1. Field of the Invention[0003]The invention generally relates to use of facial recognition technology as used in surveillance and access systems and is specifically directed to incorporation of such technology in an IP compatible, networked, comprehensive multimedia surveillance system.[0004]2. Discussion of the Prior Art[0005]My earlier patents and applications have covered various aspects of the networked multimedia surveillance system in detail. My following earlier patents and pending applications are incorporated herein by reference:Ser. No. 10 / 192,870 Filing Date: Jul. 10, 2002Title: Comprehensive Multi-Media Surveillance and Response System f...

AI Technical Summary

Benefits of technology

[0053]The basic function of the system can be enhanced by dividing the processing function processors. One or more processors perform the computationally intensive tasks of Facial Separation and Facial Signature generation, while yet another processor performs the less demanding task of database pattern matching. This yields improved system economies and flexibility. Cameras and Facial Processors may be added incrementally to the system as needed, and as is unnecessary for each Facial Processor to contain or to access the entire ‘reference’ database.

[0056]The utility of the system may be enhanced by the increased use of the networking techniques of the subject invention. In this enhancement, a group of networked processors perform the steps of Facial Separation and Facial Signature generation. The Facial Processors function as network resources, and are configured to process video from any networked camera, as required. This improves the flexibility and economics of the system. For example, during periods when a particular area is not used, Facial Processors may be diverted from analysis of that particular camera to an area of higher traffic. Also, the workload of a failed Facial Processor may be diverted to a different processor.

[0058]An additional benefit arises from the fact that IP surveillance systems often maintain an archive of stored video or images. Since this archive is generally available on the network, it is possible to use the system to search for faces in archived images, during event reconstruction. In the preferred embodiment the IP surveillance network stores captured images or video in an Image Database. Often, these images are captured only when the associated camera has detected motion within its field-of-view, thus reducing the storage requirements of the image archive platform. Since the Image Database is a generally-available network resource, it is thus possible to perform the Facial Processing on these stored images as well as on live camera video.

[0061]In a further enhancement of the invention, the IP cameras include additional processing resources, and are thereby capable of performing the Facial Processing internally. The separate Facial Processors of the previous example are thereby eliminated. This approach allows improvement of the storage efficiency of the Image Database since images may, if desired, only be stored in the Image Archive if a face is recognized by one of the cameras, or if a particular predetermined face is detected by the Facial Database.

[0062]My previous applications and patents as listed above and as incorporated by reference herein describe a surveillance system wherein the IP cameras may produce multiple video streams as the system requirements may dictate. For example, the IP cameras may produce several motion MPEG video streams with different bandwidths (for different audiences), and may additionally produce a high-resolution still frame JPEG image for storage in an image database. The system may utilize any of these video streams for facial recognition. Since the cameras are IP-based, their motion video and still frame video streams are generally available upon demand throughout the network, and either type of video may be used to drive the Facial Recognition system. The still frame images have the advantage of greater resolution, but may be generated less frequently. Motion video sources may produce useful images more often, but at a reduced resolution. This reduced resolution decreases the accuracy of the Facial Recognition process.

[0072]It is also an object and feature of the subject invention to provide improved facial recognition capability by utilizing high resolution digital camera technology to capture the image.

Problems solved by technology

Up to now, database searches were limited to textual entries (i.e., name, social security number, birth date, etc.), leaving room for error and oversight.

Images