 |
 |
|
 |
|
Perimeter Protection - An extensive area of application |
 |
 |
 Perimeter protection is an extensive area of application. It includes installations such as lighting, access barriers and enclosures as well as detection sensors for monitoring property lines. We spoke to Andreas Wolf, Product Manager Intelligent Video Surveillance with Dallmeier, about the possibility of advanced video detection.
?Mr. Wolf how can video surveillance contribute to perimeter protection?Video technology used in perimeter protection has two important functions. For one thing video recordings can be used to visually verify alarms that were for example triggered by detectors on a fence. Additionally, video can actively detect events along property lines or within a defined area.
?What is required to implement perimeter protection using video?In principal three things are needed: First you need cameras that will still provide good quality images even with little residual light in order to be able to carry out an analysis. Secondly, one of course needs a video analysis system with recording capability and thirdly a video and alarm management system. Apart from that there should be suitable locations for installing the cameras. The most important point however is careful planning with a thought-through overall concept. At the end of the day every customer has individual security demands that need to be realised.
?From the point of view of a manufacturer of video analysis systems, what needs to be taken into account when planning such a system?First of all the customer's requirements need to be clarified. For every project those requirements are very individual and decisively influence the positioning of the cameras and the processing of alarms.
?Exactly how do the cameras have to be installed?With conventional perimeter protection the cameras are installed in close vicinity to the property line. For this purpose usually masts are used that also carry the lighting. The cameras should be installed high enough so that movements towards the camera can be reliably detected. Generally, installation heights of four to six metres are okay. It is important to make sure that the mast is resilient to get judder-free images even under windy conditions. The lighting should be as far away from the camera as possible in order to have a homogenous illumination and reduce disturbances caused by insects in front of the camera.
? How far is the detection range of the camera?This question cannot be answered in general terms. The depth of detection is directly determined by the lens and limited by the definition of the smallest object. Those are decisive parameters for the analysis.
The choice of the lens determines the field of view of the camera and thus also the detection range. For the exact planning there are tools like the free View Designer by Dallmeier. We use it to determine the distances between the camera masts at the customers' sites. Thereby the decisive proportion is the size of an object that is to be detected at its maximum distance. We recommend ten per cent of the respective resolution. Therefore, with standard cameras the realistic distances between the masts are around 50 metres. We usually choose larger focal distance and pay attention to having overlapping camera views in order to avoid dead angles below the cameras.
The distance between the masts can be expanded by using higher resolutions and/or larger focal distances. However, larger focal distances always lead to larger dead angles, and higher resolutions result in more extensive disturbances. Larger focal distances and higher resolutions also don't increase the depth of field. Hence, the video image does not automatically become more conclusive and better to analyse.
However, when the interferences in the image are reduced, for example by using thermal imaging cameras, the range can well be expanded to 80 metres. This is simply due to the fact that with thermal imaging only warm objects are still detected, which results in only very few image disturbances.
?What is detected by video?In principle, a video analysis detects the moving parts in the video. Those moving objects can be people and vehicles but also branches, grass, animals or shadows. The detected objects are subject to a comprehensive validity check. Depending on the analysis system a certain number of alarms are triggered.
?What does validity check mean in this context?A validity check evaluates whether or not a given object is something that fits the alarm grid. Important criteria are the minimal size, which limits the depth of detection, and the maximal size of an object. A further criterion is the behaviour of the object i.e. how it is moving and where to. That information comes from the so-called tracking i.e. the tracing of an object within the video. For example an exact definition of a scale is in most cases not that easy, because the object size changes with the depth of the image. Therefore, perspective models are used that convert pixel data into real values. Thereby, object sizes become constant in the perspective, which means that it is possible to define that, for example, the largest object should measure 2.2 x 0.8 metres. In most video analysis systems those criteria can be defined as parameters.
?What triggers an alarm?An alarm as in the sense perimeter protection can be defined as follows: „There is a (valid) object in a defined area, within the perspectively defined dimension, for longer than a minimum period of time, with a minimal movement in a defined direction.” If those criteria are met, an alarm is triggered. That is the case for example when a person climbs over a fence which is detected by a video analysis.
?What happens to the alarm then?When an alarm is triggered, two important steps are initiated. The alarm is sent to the video server which uses it as a recording trigger. The video server stores a parameterisable time period prior to and after the alarm and associates the alarm with the alarm image. The alarm is saved in a database and can be investigated at any time.
The second step deals with forwarding the alarm to the alarm management system. The management system decides how the alarm is to be processed. With video surveillance applications the standard reaction would be to display on a monitor. Further actions depend on the demands of the customer. Basically any cameras with random recording times or with live videos can be displayed on random monitors. Often it is required the alarm is forwarded to an overriding system. Thereby, it does not matter if this is done via a hardware contact or a software message, or, as is also often the case, in the form of the video stream.
It does not matter where the video management system is installed. With a modern video server it is also possible to transmit videos using low bandwidths over wide area networks.
?Is it actually possible to identify an intruder or does video not yet have the necessary capabilities? What about using higher resolution?The term identify needs to be defined more precisely. Identifying to a video analysis system means that there is something that fits the pattern. Such an object can also be the smallest defined object, which would make it very likely that the face of the intruder will not be identifiable. In this case one could speak of indentifying. Thereby it is important to ensure that the head of the person that is to be indentified is represented large enough. This can hardly be accomplished with a perimeter protection i.e. a fence detection, particularly not if you have to deal with uncooperative objects. In order to be able to identify people additional movable cameras with higher resolution that have good and fast lenses need to be used. With a Dallmeier high-definition PTZ dome (DDZ4010/DDZ4020) it is possible to clearly identify a person at a distance of up to 40/80 metres. A person’s characteristics can still be well recognised at distances of up to 80/160 metres. In the case of an alarm, the dome cameras are moved to a defined preset in which the intruding object was alarmed, for example by fence detection. From that moment on the object is traced with the dome and thus recorded over a long period of time.
?That all sounds really good but does it actually work, too?Video analysis is not like a light bulb that either gives light or doesn’t. Video analysis rather varies between being `problematic´ and being `really good´. Assessing how good video analysis works in a specific application requires expert knowledge. The most important step in realising a satisfying perimeter protection is and remains professional planning. Professional installers should rely on support by the manufacturers, while the end customer needs to be thoroughly consulted regarding his demands. A sound solution can only be worked out in close exchange and communication with the customer. We offer assistance to this end in the form of questionnaires which clarify the most important points. However, even with the best planning and implementation a perfect perimeter protection will never be achieved. There will always be false alarms caused by weather, animals and objects within the field of view of the camera. Any measures taken to reduce the number of false alarms can only work with a good installation.
?What do professional installers need in order to install and configure a Dallmeier perimeter surveillance?With the development of our analysis system, the DVS 1600, we have put special emphasis on a consistent and intuitive user concept. We have developed a configuration assistant that guides the user through the necessary steps. That way nothing can be forgotten. All working steps are equipped with an online help, which means that manuals are no longer necessary. We view this as an important step towards making video analysis more attractive again.
?Can other sensors be integrated?Of course! Dallmeier systems are open! We can integrate other sensors via hardware contact or software message. That capability was already integrated in the first recorder versions.
?What other equipment is needed besides the DVS 1600?The alarm message manager PGuard is needed to process the alarms, for visualisation the video management software PView is required.
?What about IP solutions?The trend with new installations clearly goes towards full IP. Dallmeier offers a wide range of IP cameras, among them the HD cameras I mentioned before. When the DVS 1600 is used for video analysis as part of perimeter protection applications it doesn't matter at all which form of video transmission is chosen. IP, analogue and also hybrid solutions are possible.
?How much servicing and maintenance does such a system require?As far as our experience goes similar problems repeatedly arise regarding perimeter protection. On the one hand it's animals, mostly rabbits, that trigger false alarms. However, such false alarms can be consequently suppressed by using the integrated perspective that I earlier described, given that it is correctly equipped. On the other hand, vegetation that grows over the course of a year causes problems. So, even a very well planned installation may require some fine-tuning from time to time.
A simple and effective maintenance and service of the system is thus very important. That is why we have realised the analysis via web interface (port 80). Using our configuration assistant it is possible to finetune existing analyses. This can be done conveniently via a standard internet connection, without the need for costly on-site tasks. The internet connection can for instance be realised with a short-term UMTS connection. Of course, VPN connections are also possible. A perimeter surveillance system that uses a DVS 1600 can thus be conveniently managed by remote control. Correspondingly, the possibility to access event and live videos using Dallmeier PRemote via DSL and UMTS networks is an integrated highlight of the Dallmeier technology. [www.dallmeier.com]
|
| |
|
|
 |
|
 |
|
|
|
| |
 |
|
|
 |
| |
|
|
|
