Introduction

The false alarm track analysis system is designed to help identify and eliminate false alarm tracks, such as those generated from debris, reflections, or swaying trees. Filter models have been created using a set of user identified false alarm examples. These models calculate a combined score for a track being a false alarm.

For more information about using False Alarm Analysis, please navigate here: Using False Alarm Analysis.

Contents

False Alarm Analysis

1. To enable this feature, select the Enabled check box:
   
   

   Open

   

    

Model Settings: Enable/Disable the False Alarm Analysis feature.

Data Source:

False Alarm Tracks: This will use the tracks marked as False Alarm.

Model Files: This option is for testing purposes; it allows you to use model files from another system.

Models

Location Cluster: Quite often, false alarms will occur in a certain area and this is usually triggered by the movement of debris or plant matter. This filter scores a track based on whether it took place at, or near to a location where other false alarm examples have happened. This filter will output a higher score if a live track is currently at such an area, and will be reduced, depending on the distance that the live track is away from the centre of the location where the false alarms have occurred.

Time: False alarms frequently occur at the same time each day, usually due to environmental factors which trigger the false detection. This filter scores a track based on whether it occurred in the same hour time period where other false alarm example tracks have taken place. It outputs the highest possible score if the live track happened in exactly the same hour window as a false alarm example track, and diminishes by half, for every hour of time difference between the false alarm example track and the live track.

Area Covered: We have found that many false alarm tracks do not travel over a large area; they tend to ‘jump’ about in a very small physical area. So, this filter scores a track based on whether it has a similar area covered to that of other false alarm tracks. The filter score diminishes as the area covered by the live track gets larger than that of the false alarm example tracks.

Strength Weight Cluster: Many false alarm tracks have a very low strength and weight when they are caused by noise, yet real objects often have a much higher values. This filter scores a track based on whether it has a similar strength and weight to an example false alarm track. The maximum score is achieved if the strength and weight of a live track exactly matches that of a false alarm example, and scores diminish as the difference between these values grows.

Direction Change: As many false alarm tracks tend to ‘jump’ about in the same area, they register a very large direction change. Genuine tracks do not do this, and generally have a very constant direction. Therefore, this filter scores a track based on whether it has a direction change that highly varies between sightings. It gets the maximum score if it varies by an angle of more than, or equal to the standard deviation of the direction change for all false alarm example tracks. The score diminishes as the direction change of the live track decreases.

Sightings Count: Many false alarm tracks do not last for a long time, and hence have few sightings. This filter scores a live track proportionally to the number of sightings seen for the track. For fewer sightings, the track will get the highest score of being a false alarm. Conversely, for numerous sightings, the live track will get the lowest possible score of being a false alarm.

By default the Location Cluster, Area Covered, Direction Change and Sightings Count are enabled.

If you wish to update a set of filters:

1. The Regenerate option will open the following dialogue. Click OK to proceed:
   
   

   Open

   

Related Information