IPVMU Certified | 02/04/13 04:53pm
Hello, The scope of your problem is key. Are you looking for a system that records all take-offs and landings? If so, it might get complex and is subject to a host of follow-on questions, first being: what does 'monitor' entail? detection? identification?
I worked on a project involving auto-tracking with PTZ cameras at an airport - the purpose was to record aircraft ID numbers before takeoff. See the update here.
In that situation, aircraft movement was restricted to the same 'box' for every takeoff, and we tightly restricted camera movement within that box. The result was good, but we weren't working with the 'half-kilometer' ranges (the huge FoV to monitor) you describe.
There are numerous PTZ systems that can track a moving object, although it would probably have to be a more advanced one to track at those speeds. Auto-tracking is something of a gimmick in most cases, IMO, because they don't have the ability to intelligently determine WHICH moving objects to track when there are multiples, but in this case, there's probably normally going to be only the one object in the FOV, so that shouldn't be an issue.
The other thing you could do, depending on where the camera is to be located, is use a wide-angle, even hemispheric camera to view the entire runway - depends on what level of detail they want of the entire thing. Say, a 180-degree camera placed sufficiently off the side of the runway around a midpoint?
Brian... yes, it involves tracking (recording) every landing/takeoff. We can use two cameras one for landing and other for takeoff. Basically the airport want to track the movement by ID# marked on the aircrafts. So higher image quality is needed for identification. Half kilometer range is to detect the object, but identification should be possible at closer ranges.
Matt, technically it is going to be only one object in the FOV. However, we still need to record the ID# of the aircraft.
How about one camera (either tracking PTZ or hemispheric MP) to capture the takeoff and landing, and another focused on a specific point that the planes must pass to get the ID?
Brian's experience may prove otherwise, but personally, I don't think I'd try to force one camera to do both tasks (ie. takeoffs *and* ID, or landings *and* ID).
To me, it's a similar idea to what we used to do for gas stations when they were concerned with capturing license plates (before a gas-prepay law was brought in and gas-and-dash events dropped off): a couple or a few wide-view cameras for general site activity, and then a tight shot on each entrance and exit to get clear shots of license plates.
Brian, few questions on the airport project -
1. Which Axis PTZ cameras were used?
2. Do you think the auto tracking can be employed as/when the aircraft lands at high speed?
3. How far was the camera installed from the designated "60 foot box"?
4. Any specific auto-zoom lens was used?
5. How about night runs?
We have a strikingly similar requirements. We are planning to test out Axis 6035-E PTZ for the runway. For the 1/2 kilometer descent and takeoff, we plan to test a fixed Axis P1347-E camera.
IPVMU Certified | 02/06/13 03:33pm
1. Q6032-E (The 6035 was not released at that time.)
2. I do not think that trying to track a fast moving object will work well. It is hard to translate airspeed/groundspeed to PTZ movement, but I would estimate movement greater than 60 degress/second is difficult. However, that's just a really rough guess in my mind, and it should be tested.
3. The camera(s) were installed about 200-400 feet away from the staging 'box'. Cameras were installed on hangar buildings physically located on the apron.
4. The lensing on the Q6032 was not something that could be changed. We used no special auto-zoom settings in the application, just defaults.
5. The solution was used at night, but the apron where the 'box' was located was BRILLIANTLY lit at night. I mean, there must have been 5000 - 7000 lux worth of illumination over that whole area, so nightime wasn't a big factor.
And to clarify: This application was not used to verify takeoffs/landings at speed. It was used to record airframe ID number BEFORE they entered the taxiway for takeoff, with planes pausing in the box for permission to begin. Landings were an event that happened outside what these cameras were used for.
IPVMU Certified | 02/07/13 01:55pm
SightLogix comes to my mind for PTZ tracking. Maybe they could modify their analytics for your purpose to look for and track planes. (I'm sure there would be some significant development costs.)
[Editor's Note: Commenter is from Sightlogix]
SightLogix would work for tracking the aircraft while on, or very close to the ground. This would be done with one or more fixed SightSensor thermal cameras covering the area of the runway with the number of cameras needed dependent on the camera angle and mounting position options available. The PTZ tracking camera would be steered by a SightTracker unit that translates the fixed position SightSensor(s) position information into azimuth, tilt angle and zoom positioning for the PTZ. We have a layout tool that can easily approximate the layout should Pushkar wish to pursue a possible solution using the SightLogix system. Contact John Horton at 609-951-0008 x109.
IPVMU Certified | 02/08/13 02:25pm
In lieu of thermal, you might also consider ground-based radar like SpotterRF. However, the speed of a target like a plane may be trouble, and field tests are sure to be a good idea.
SpotterRF integration with PTZs or VMS is likely to require additional integration cost.
Thanks Guys...I'll discuss the possible options with the client.
Jim, thanks for sharing John's information. I'll get in touch with him, if the client likes the SightLogix option.
If the aircraft have and the airport uses IFF, they might be able to provide you a feed of (maybe) reported altitude and range/bearing from the airport secondary surveillance radar. If the aircraft use ADS-B, you can track, filter, and predict aircraft position. Does the airport have any local primary airport surveillance radar that might provide this sort of rough localization which (I would imagine) would be helpful getting the PTZ pointed in the ballpark, but would probably still require video tracking to keep it centered in the field of view?
One way that such a digital feed might be palatable to the IT/IA guys is, it can be set up as a one-way spigot, avoiding any possibility of interaction with safety of flight systems.