Blog

Working characteristics of thermal cameras are very different from those of conventional cameras. The image of a thermal camera is monochrome and is a representation of relative temperature of objects.

A person or an animal is visible in a thermal image in as much as their temperature is different from the temperature of the background. The contrast in the thermal image can be very different depending on time of day or weather conditions. For example, in an outdoor scene of a cold night, a pedestrian will appear white of light color on a black or dark background. However, in the same scene during a sunny day, the pedestrian walking on hot asphalt will appear black (or dark) on a white or light grey background.

The OneTrack intelligent detection functions work very well with thermal cameras since all is needed for a good detection is a good contrast between objects and the background. The thermal sensitivity becomes especially useful for outdoor surveillance under very low light conditions.

For alarm central monitoring, OneTrack intelligent detection coupled with thermal camera can result in excellent sensitivity with very low false detection. In addition to sending messages to a control stations, alarms can be sent email or to a private monitoring station based on VideoWave’s CMS software.

In a general context, let’s say it is desired to monitor an outdoor area with a PTZ camera and a fixe camera. If the fixe camera is used to detect objects and the PTZ to track them, then in night (low light) condition a conventional fixe camera will not have the excellent contrasted image of the objects that a thermal camera provides and thus the tracking of objects will not be as good as with a thermal camera.

Until recently, the high cost of thermal cameras restricted their usage in many commercial applications. However, new models have been brought to market recently that expand the possibilities for these applications.

We have tested several models from FLIR and have conclude that the new models with very low resolution (80, 45) still deliver good results, when used as detection, not as view image.

Path zone – How it works?

Configuring this zone is done on the fixed camera. The zone is drawn along with the path direction. We then specify the exact Pan/Tilt/Zoom position of the PTZ, along with the fixed delay of time to stay on this zone.

When configured, the PTZ camera performs normal tracking operations when nothing special happens into this zone. Once the fixed camera detects an object (vehicle) into this zone, the PTZ is called in priority to move to the previously specified Pan/Tilt/Zoom position. The PTZ will stay in that position for the specified delay. By default, this delay is around 10 seconds. This delay is normally enough to grab a high quality image that allows to read the licence plate. After this period of time, the PTZ goes back to it’s normal tracking operations.

This function is really useful for outside security systems by providing a high level video analytic function without the requirement of dedicated cameras. Incidentally, this function can also count the number of detections in this path zone (ex: number of cars that leaved in a day).

The characteristic of thermal cameras are very different than conventional cameras: even if the thermal camera gives a lot less details than a natural image, the thermal image filters a lot of noise and reveals a lot of information that would be hard to detect, mostly at night.

These special characteristics makes it efficient to detect persons and vehicles at night, in low light and even in full sunlight. The OneTrack system can perform efficiently object detection on those thermal cameras, IP or analog.

Testing performed with thermal cameras with new object detection functions of the OneTrack allowed to detect a pedestrian at night at a distance of around 400 feet, with a minimum of false detections.

AutoTrack camera or MultiTrack system?

PTZ tracking systems essentially works in 2 modes: Autotrack or Multitrack. In Autotrack mode, the PTZ camera detects the motion in the image and positions itself by trying to center the detected object. Of course, when the PTZ is oriented in one direction, it cannot detect what’s going on in the other direction. To perform detection in a 360 deg angle, a tour is required. During the tour, the camera performs the video detection at each step and launches a tracking if required. However, performing a tour requires some time, so it’s possible to miss a fast intrusion.

In Multitrack mode, the video detection is performed on fixed cameras associated to the PTZ. When an object is detected on a fixed camera, the controller positions the PTZ directly on the detected object. The fixed cameras can be positioned to cover the entire zone to monitor. If many objects are detected, the PTZ can jump from one object to the other at a fast speed. Even more, since the detection is done by the fixed cameras, the intelligent video analyzer can classify the objects (car, pedestrian) and filter false objects like reflections on puddles, waving flags, etc. In Autotrack mode, since the PTZ is always moving, there is no background available to perform this kind of classification.

Until recently, connecting IP video surveillance equipment of different manufacturers was problematic. Each brand used their own proprietary communication protocol.

An international forum wants to change this situation by defining a standard specification for connecting those equipment. There are multiple advantages of the ONVIF standard, including:

• Easier connection between the various security equipment by using a unified protocol;
• Greater flexibility when choosing the system components by reducing compatibility requirements.

Even if the ONVIF specification is quite recent, this standard has all the potential of bringing a lot to the security industry.