Trusting An Alarm Is As Important As Answering It, So That's Why We Eliminate The False Ones

Guarding Against Nusance Alarms
Nuisance Alarms in the detection of wildfires can be caused by a number of reasons. Some of the most prominent are:

• Cloud Movement
• Cloud Shadows on the ground
• Vibration of poles, trees, leaves etc
• Camera vibration
• Sun reflection on bright surfaces

The WDS software provides multiple tools for the operator to minimize and in most cases drive nuisance alarms to zero by using the following approaches:

The drive to reduce or eliminate nuisance alarms begins in the core technology of the system. The WDS system evaluates a continuing video stream at each set point while most other systems evaluate a few still images. This provides the WDS system with significantly greater information with which to reach a decision.
The second key design aspect is that the WDS system uses a two stage evaluation process.   When key detection parameters are met, the system issues a visual (non-audible) warning that an actionable event is possible.   It continues its evaluation to finally reject or accept the event causing an alarm. This two stage process is possible due to the fact that the system is receiving a video stream rather than still images.
During set up or at anytime during operation, the sensitivity of the system can be adjusted. The target is to reduce the sensitivity to the lowest possible level that still insures a 100% detection result. The company recommends a default setting that can be adjusted based on the requirements at each location.
Camera positioning cannot always eliminate sources of nuisance alarms such as industrial locations, moving poles or structures, or horizon areas containing clouds. The WDS program, can exclude each of these sources through its "Region of Interest" option which allows the operator the ability to draw on the screen those areas that the system should consider as a source for actual events.
Extend the motion evaluation time by selecting an option (Return to Alarm Preset) that will hold the image stream evaluation in memory while the system moves to the next parked position, evaluates that parked position, then returns to again analyze the dynamics of the held video evaluation. This allows for greater expansion of the potential smoke moving region.
Select the addition of a support vector machine (SVM) decision making algorithm set that has been trained with a massive data set beyond basic smoke characteristics.
The system provides two options in the event of camera shake installation