Design Perimeter Monitoring With CCTV Only

Can you clarify a few things? When you say 'monitoring distance from the fence is 60 meters' does that mean you are going to mount cameras 60 meters away from the fence?

Do you want to use analytics? Do you have any lighting on the perimeter? What do you want to 'see'? Just detect people or do you want to be able to see personal details?

They want to identify intrusion based on human eyesight or they want the intrusion detection to be done automatically by video analytics?

Secondly, the distance away from the camera is less important than the width of the FoV. How wide of a FoV are you expecting or do you need? Background reference - Field of View tutorial.

Hi John,

It will be Manual eyesight detection.

Also There are cameras placed every 60 meters throughout the perimeter to avoid gaps , around 166 cameras.

Can you suggest what best can be achieved with a pelco fixed IX camera?.. Shall i recommend Tele focus in this case..

So do you want to "target objects 60m away from the fence" as in outside/beyond the fenceline or do you want to put each camera 60 meters away from each other, looking straight down the fenceline? I assume you mean the latter but want to make sure.

Hi John, the latter one..

10 kilometers of fencing, and you want to have 1 camera cover every .06 kilometers? How many guards will you have watching this ~166 camera system?

I really think you should consider using some sort of video analytics for first-level processing of the perimeter, unless you have ruled that out for some reason. You could cover 100 meters *easily* with an optical camera, or 500+ meters using a thermal camera. It would be a significant reduction in equipment and in man power needed to monitor the system.

To answer your original question, the best design would be dependent on the overall scene. A ideal case is that the fence line is cleared of vegetation and has very little activity, allowing the guards to key in on any visual motion in the scene. As the visual complexity of the scene goes up, so does the requirement for the guards to concentrate more on the video, which may require the use of more cameras and/or the use of more video wall equipment in order to display the video feeds large enough for any motion to register.

This sounds like the kind of application that is known for causing human guards to zone out and miss things.

VMD for outdoor perimeter protection is likely to be a disaster. Granted, it will be cheap but you'll get the end user fired :)

You are probably better off with thermal as long as you simply want to detect people, not see personal details. It will likely be less expensive too considering the longer range you can achieve. For example, see this thermal vs color test (thermal's advantage at night is massive even against top color cameras and with a more telephoto lens, you can even see farther). Also, see our thermal resolution shootout, testing up to 500 meters.

I'd second looking at thermal. We did a setup a few years ago along these same lines - not quite 10km of fenceline, but a two-square-block mechanical yard watched by thermal cameras on a tower in the middle: the cameras feed to an analytics system which then has output tied to an alarm panel to alert a monitoring station if it detects a person within the yard perimeter. There's also a Pelco Esprit PTZ on the tower so the monitoring station can look in and see what's going on.

In my experience, you can get good coverage to ~100 meters using our (VideoIQ) HD cameras and some form of illumination (typically IR) for night time operation.

The cost-per-foot for the equipment comes out pretty close between thermal and optical setups, all things considered. Thermal will see further at night, and reduce the total amount of units you have to install. Optical will provide a color image for the majority of the day, which if intrusions are likely to happen at any time during daylight hours, it's nice to have a better view of the person intruding, which you wouldn't get with thermal.

Technically, you could push an optical camera to cover 700' comfortably, but you run too much risk (IMO) with atmospheric interference in some climates and conditions once you start to go beyond ~400' (which is why I use 100 meters as a conservative estimate, based on practice). Lighting will also become more expensive and more of a challenge at those distances (maintaing even lighting across the range, etc.).

For your application you wouldn't want any kind of motion based detection, you need something that has bona-fide object classification analytics (call me biased).

You might also look into other technologies and maintain interface between VMS and intruder detection.

You might consider some of fence alert systems such as FFT using optical fibre and interface with VMS.

Hai Undisclosed,

Why can't you use PIDS (Perimeter Intrusion Detection System) integrated with PTZ Cameras, which makes more sense instead of depending on Guard... Also we can reduce number of cameras, say for example every 75Meter radious u can use 1 PTZ Cameras integrated with 4 or 5 PIDS zones....

"With sarix 720p I am only able to cover 60m with 39mm lens and the human at 60 m falls under identification range."

I don't think you want to do that...

At 60M, you'll end up with a FOV about 7M wide, and it will be considerably narrower than that for most of the area between the camera and the end point. There is too high of a risk that a person passes through the FOV undected due to a very short appearance time. I think what jvsg was telling you there based on calling it "identification" was that you'd have enough pixel density on target to recognize the person. In all likelyhood this won't end up being true in the actual deployment since you won't get good facial shots anyway.

Also, virtual lines are usually my last recommendation for something like this. Too much chance again of the target being missed because they were not being optimally tracked while crossing the line. We usually just recommend a region of interest (ROI) covering the "secure" side of the fence. If the area outside the fence is generally low activity you can extend the ROI outside the fence to get an extra few seconds of early warning so the system trips when someone approaches the fence.

Hi,

Abertura Photovotaic solar plant chooses FLIR Systems for perimeter security - YouTube

YouTube

anyway, if you consider worst meteorological condition: heavy fog, heavy rain or snow, working distance for analitics

using normal cameras could be 40-50m and thermal up to 100m.

A good analytics (IVA) you can find integrated on Bosch cameras

YouTube

Manual eyesight detection for such a large perimeter = forget about it.

yup thanks for the insight. I am strongly recommending analytics as well in my bid

I believe fence/peremiter detection via alalytics is a SightLogix specialty. I think they are about $10k per camera. If your fencing is mostly straight fence sections then this might be a good fit, if you have lots of corners and turns then probably not.

I have demoed this product at my facility. The analytics and detection are excellent. The PTZ integration is decent (not great). The GIS is also just decent. It has been at least 2 years, so the product may have improved. I didn't buy because we managed to convince the the FEDS we didn't qualify as a TIER1 risk facility (whew!). This was the leading contender though. DVTEL was a close second.

You might want to have a look at DVTel IOImage cameras which comes built-in with Video Analytics.

AVT234 VMD Fence Sensor which integrates with any camera and provides serial alerts to any DVR/NVR, VMS, Annunciator Display system. 8 to 16 channels per device

PLE's AVT234© has been recommended for USAF PL-1 certification for Video Motion Detection Software. It is supplied on a tested, ruggedized, 2U black box server.

AVT234© delivers superior indoor and outdoor Video Motion Detection (VMD) that provides consistently accurate alerts and almost zero false positive alarms in all weather conditions.

Hello,

For an Intrusion detection System on straight fence lines what do you feel is the better solution?

Assumptions; cameras will be used for verification in both cases; cost is not a factor; that vegetation is clear 10' on either side of the fence; main goal is to reduce false alarms.

Which solution would provide the best fence line detection with the least false alarms? 

1. Use analytics on Thermal cameras (assume correct spacing) 

2. A fiber based detection system (Senstar or FFT)

Thank you in advance