Long Term Clutter Map
In normal tracking the clutter map (long term average) has to update reasonably fast. This is to ensure normal clutter, such as trees and long grass, does not generate any false alarms. This works well with moving targets, such as normal traffic, however the downside is that tracks will be absorbed into the clutter map when they become stationary. This can happen in a matter of seconds which means stationary objects, such as debris, can be missed.
By having a much slower updating clutter map, stationary targets are observed for longer and the chances of seeing small, static objects are much improved. This approach does increase the probability of false alarms however we use the second key process, multi-scan analysis, to focus on just the targets we want to detect.
Multi-Scan Analysis
This process differs from the typical radar processing because it looks at data over several scans. Essentially it looks for signals which are greater than the specified signal threshold over a number of rotations. Depending on how many times it see a target exceed the threshold, the process will generate an artificial signal based on the total number of detections. This provides a high level of confidence that the object being examined is in fact a genuine piece of debris. The process does delay the creation of the track but typically this only by a few seconds.
Classification
The debris channel is configured to always report any target as debris. This sounds logical, but also consider that any stationary object, from small to large, could be reported as debris. This would include stopped vehicles or people standing still. This is a minor disadvantage consider the overall improvement in performance and futures versions of this feature will include more intelligent classification to correctly distinguish between small and large objects.