At the moment, IP and networked video is primarily being used to solve the traditional CCTV problems of video transmission and storage. However, the industry is slowly realising that computer technology can provide many more opportunities for the end user.
Key areas where IP has been proving beneficial with CCTV surveillance are remote management, Automatic Number Plate Recognition (ANPR), facial recognition, automatic incident detection and intelligent image retrieval. The technology used by the CCTV cameras also needs to be explored.
IP and remote management
Together with a secure WAN and firewall, IP technology brings the ability to manage security and surveillance systems from a central or remote location that can be (literally) anywhere. As little as one or a few operators can control and manage all locations, reducing the cost of security personnel on site. By operating remotely, this also provides a more secure place in which the operator can work.
IP technology enables Control Rooms to back each other up, providing additional security to the systems themselves. Runnymede Borough Council and St Peter's Hospital is a good example, whereby the Council enjoys the ability to link up with the cameras monitoring the hospital and provide a back-up should the need arise.
Security officers and CCTV operators should not be tied to the Control Room at all times. With IP technology, they needn't be. Each officer can be equipped with a standard, handheld pocket PC, such as the Compaq iPAQ, and secure IP network link to the CCTV system. From the pocket PC, the security officer may then select a particular camera on the CCTV system and view live video, open or close locked doors and talk directly to the police from his or her mobile 'phone – all without being tied to the Control Room.
This engenders truly mobile CCTV control to the CCTV operators and security officers wherever they are within the building or location being monitored.
Importance of trigger events
IP technology brings with it the ability to create trigger events. These are actions pre-determined by the security manager so that when that action takes place, the system is automatically triggered to take appropriate steps. Actions such as a door opening/closing, movement within an area or changes in, say, room brightness can all act as trigger events.
Using IP, the operator is able to view only those cameras where trigger events have taken place. If someone presses an intercom to access a building, an event is triggered and the operator duly alerted. The camera closest to that event automatically focuses on the area, allowing the operator to view the current situation. He or she can then speak to the person who pressed the intercom and decide whether or not to grant access.
Cameras will also automatically start to record a trigger event, avoiding the possibility of the operator failing to initiate this procedure. In addition, IP technology allows the operator to access what has already been recorded and view footage. Additional triggers may be set up after the event to pick up on any further evidence that's available. For example, a would-be thief in a shop can have his activities tracked by the user creating additional triggers based on the clothing he or she was wearing. By doing this it's possible to see which shops the thief went into during their 'spree'.
Automatic Number Plate Recognition
Automatic Number Plate Recognition (ANPR) is an image-processing technology that's used to identify vehicles by their license plates. It's currently adopted in various security and traffic applications. Using IP technology, ANPR can operate either on a stand-alone basis or as part of an integrated CCTV system. Typically, ANPR systems will identify vehicles travelling at speeds of up to 120 mph, delivering image accuracies in excess of 95%.
ANPR affords the ability to automatically check vehicles against a variety of databases, including the Police National Computer (PNC), DVLA and other local databases. Once a license plate has been read, an ANPR system will generally be used to perform one of three functions: 'data logging', 'blacklist matching' or 'white-list matching'.
'Data logging' records the details of every identified vehicle to a database, which can then be browsed or searched on a number of data fields such as vehicle nationality or state, an indication of the accuracy of the recognition process, the vehicle's location or the time and date that the vehicle was logged. This can also include 'wild card' searches against the registration number. This is often used by the police when a vehicle is seen behaving suspiciously, or is pulled over.
'Blacklist matching' cross-references every identified vehicle against one or more 'blacklists' of suspicious vehicles. If a match is found then the operator is immediately alerted and an alarm sent to the police or some other external system.
By way of an example, West Kent Police used blacklist matching in Operation Scallywag, a recent project involving the use of ANPR to check vehicles on one of Kent's major roadways. By checking each registration plate against various databases, 'rogue' vehicles were pulled over for further checking – resulting in over 30 offences being detected in a six-hour period.
'White-list matching' matches the number plate of each detected vehicle against one or more 'white lists' of known vehicles. If a match is found, full vehicle and driver details pop up, from which a resulting action can be taken. For example, an employer may raise a barrier to allow entry to the company car park, or a fee may be charged to a customer – as is the case with Congestion Charging in central London.
Security officers and CCTV operators should not be tied to the Control Room at all times. With IP technology, they needn’t be. Each officer may now be equipped with a standard, handheld pocket PC and secure IP network link to the CCTV system
It's only within the last year or so that ANPR has been widely used in CCTV schemes, and it has made an immediate impact. Northampton Police officers have tested ANPR technology in the town centre and obtained excellent results. In a period of one month, vehicle crime fell by 13.2%. Eleven stolen vehicles were recovered, 42 arrests made and the police prosecuted 204 separate traffic offences.
Facial recognition technology generates automated non-invasive face surveillance and identification that can yield instant results. It automatically locates multiple faces in a scene, identifies the unique characteristics of each face and check through millions of records for possible matches. This can help to identify known criminals or terrorists, or may be used to quickly pinpoint staff members or VIP guests at events.
Combining facial recognition technology with commonly-used handheld PCs enables security officers to be mobile and patrol premises while maintaining a watch over who is entering and/or leaving a given building. This gives the officer a far more effective and efficient means of securing a building against intruders or any other unwelcome visitors.
If the recognition system detects a possible match against the database of known faces then it will instantly send an alarm to the operator, thus alerting him or her with pictures of the face that has been spotted and the corresponding matching face in the database. The operator then verifies or rejects the match, and provides additional feedback as they locate the target and apprehend that target. If several security officers are on patrol, it's also likely that one of them will be in a better position to quickly reach the target.
Identifying threats to security
CCTV surveillance traditionally relies on operators checking numerous monitors simultaneously, and then picking up on any momentary incident that may occur. This is a repetitive manual task relying on the operator to spot these incidents, which can easily be missed. IP technology enables this monitoring to be automated, bringing uninterrupted 24x7 surveillance of not just every monitor in the Control Room, but every camera on the network. It also reduces the chances of an incident being overlooked.
Using IP technology, the computer can isolate different objects viewed by a camera, such as people, vehicles and colours, etc and quickly detect any unusual occurrences or incidents. An alert is then sent to the operator advising them of the presence of a person, people or a vehicle, for example, so that the necessary action can be taken.
It is possible to programme the computer to learn the process and sequence of events. For example, it's usually the case that when a door is opened someone then enters or leaves, and then the door closes. As a result, the computer can quickly identify abnormal behaviour, such as someone running out of a shop, or a car parked on a double yellow line.
However, many problems are only identified as the incidents occur, rather than them being seen as a threat during the build-up period to that incident. For example, a pickpocket is often only identified when seen running away from their victim. Yet there is a distinctive pattern adopted by groups of pickpockets as they go in search of their unsuspecting prey. Similarly, a terrorist attack is well planned in advance, and will include the 'casing' of the target area by people and vehicles many times before the attack actually takes place.
Only with the use of IP technology – together with good databases and data storage/retrieval systems – can we now start to pre-empt these activities and look at the prevention of an incident rather than acting only after it has already taken place.
Intelligent image retrieval
As the number of installed surveillance cameras and the use of IP technology increases, so the CCTV industry is facing the prospect of large multimedia archives whereupon it may be very difficult to locate specific content.
To realise the full benefits of this wealth of data, it's necessary to be able to automatically highlight events of interest to the operator in real-time. We also need to make it possible to quickly identify and retrieve content that meets particular criteria.
Most CCTV footage goes unwatched, with no-one caring what's on the tape or disk. When an incident occurs, a small piece of footage may contain critical evidence that then results in the police taking thousands of hours to locate. This is obviously a laborious task that is not only costly but often inaccurate.
With the use of IP technology, we can automatically tag CCTV recordings that describe the video content, and locate and retrieve particular events very quickly indeed. While there is no foolproof means of reaching the required piece of footage instantly, it does at least highlight those areas of footage that should be investigated in the first instance.
Working efficiently and effectively
In conclusion, the introduction of IP technology and networked CCTV has made it possible to improve many areas of the surveillance process. By automating processes many major benefits have been achieved which were previously impossible when using traditional analogue systems. Control Room operators and security officers can now work more efficiently and effectively, and with a far higher degree of accuracy.
Source
SMT
Postscript
Sally Weekes is marketing manager at Petards Vision (www.petardsvision.com)
No comments yet