provided by: 
The market for IP video surveillance systems (IPVS) has grown exponentially. Network cameras can be easily installed wherever there is a wired or wireless network. Organizations are deploying an ever-increasing number of cameras, which can lead to progressively larger systems, as well as challenges and costs in monitoring all these streams of video content.
The need for efficient security and surveillance calls for an entirely new set of control tools akin to Internet protocol (IP) surveillance. Content analytics enable the mining of critical information from vast streams of live and recorded content.
The IPVS Foundation
The first IPVS systems employed digital video recorders (DVR), devices that served as the cornerstones of surveillance. These earlier generations of IPVS systems were proprietary, making it hard to mix and match components from different vendors. The industry is moving to open standards and architectures. "Closed" digital video recorders are now being replaced by more flexible and open network video recorders (NVRs), designed to run on computer servers.
NVRs use standard computer components, operating systems and IP technology to allow full control and management of video surveillance cameras over the LAN, WAN and Internet. This has revolutionized the industry by taking the concept of digital video recording and making it an open-based network software application.
The drive to open, software-driven IPVS has presented challenges and opportunities to the industry and security users. Systems are increasingly being built using commercial, off-the-shelf computing components, driving down costs. IPVS systems can be installed on top of existing wired or wireless networking infrastructures, eliminating the need to run and maintain coaxial video cabling.
Further, open IPVS systems are now often being integrated with other digital security systems such as alarm panels, access control systems, point of sale and biometrics, making it possible to put in place comprehensive, intelligent physical security solutions that transcend simple video monitoring.
These very advances have also presented challenges to security users because burgeoning cameras mean more video streams to monitor and more diverse system inputs to manage.
Advancements in video intelligence and content analytics address these problems. Systems that employ the latest in content analytics can present camera output that has been triggered by specific events. Any incidents that match predefined criteria automatically alert the operator and stream video from that camera, reducing the need to watch inconsequential images. The same number of personnel who managed four cameras can now manage as many as 100. As such these systems not only help intelligently manage growing systems and video content, but do so in a way that actually ensures greater security at a reduced cost.
A Content Analytics Overview
To better understand content analytics, think of inputs, outputs and the intelligence that sits in-between. Inputs are the external stimuli detected by video surveillance systems. These include the monitored fields of view covered by an installation's cameras and electronic signals from other security systems.
The inputs are sent over the IPVS network and evaluated by the content analytics module, which compares these signals for matches against predefined threat signatures.
When integrated with IPVS systems and combined with advanced video intelligence capabilities, the technology automates the process of detecting events that indicate potential security threats, and alerts to appropriate personnel. The IPVS system generates outputs that identify for the appropriate security staff the relevant images and in some cases couples these with audible alarms.
Types and Examples of Content Analytics
There are three main types of content analytics: object detection, biometrics, and optical character recognition (OCR). Facial and iris recognition typify biometrics content analytics. These technologies are generally used to uniquely identify people for access control applications.
OCR content analytics can be used to scan license plates, at low or high speeds, for matches against databases of suspect vehicles.
Object detection systems look for anomalies that can portend threats: crowding, motion, lack of motion/loitering, object removal or "tailgating." Tailgating detects situations where individuals are following too closely to one another — think of a pickpocket and its mark, or of a person entering a bank ATM vestibule closely on the heels of another.
Object detection provides rich possibilities for automated threat detection. The most sophisticated systems make it possible to define myriad conditions that imply threats. For instance, areas can be marked and thresholds set based on elapsed time and object quantities, shapes and sizes. Is there motion where all should be still? How long has that box been sitting there by itself in the airport terminal?
Integrated systems employing video intelligence go beyond content analytics. They process and evaluate various inputs from different components of the security system and detect a wider range of threats. The integrated system equipped with video intelligence results in a security platform that enables logical decision-making based on events and pattern matching.
For example, a system could be used to make sure that the number of people entering the protected premises is the same as the number of people swiping their security cards. The video system can detect tailgating behavior, count people and compare this with signals from the access control system triggered every time someone swipes a card. As such, it can tell if a person without an access card is tailgating.
The Future is Here
Surveillance systems are often used for forensics, to record what happened at a crime scene and identify perpetrators. But by adding intelligence, the IPVS systems' network of "eyes and ears" can increasingly be used to notify emergency responders about crimes in progress.
In the public surveillance of high crime areas, IPVS systems equipped with content analytics and video intelligence can automatically detect threats wherever cameras are stationed. When these systems work with a mobile wireless network and are coupled with mapping software and global positioning systems in patrol cars, the alerts can automatically be routed to the closest officers. The live video stream of the trouble spot (and surrounding cameras) can be "pushed" to the police vehicle to give the officers the information needed to intercede and prevent the crime or capture the perpetrators.
The ability to intelligently deliver the right security information to the right personnel at the right time — or to have your video monitoring wall remain blank until specific 'events' populate selected monitors and draw attention to just the cameras and video streams that show suspicious patterns changes the game in security, streamlining operations and reducing costs while focusing attention and resources on real threats.
Gadi Piran is president and CTO of On-Net Surveillance Solutions Inc. (OnSSI).
author: By Gadi Piran