Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

Introduction

Overview


Panel
borderColorsilver
borderWidth1

On this page:

Table of Contents
maxLevel3
indent10px

Automatic Track Initiation

In order to calculate how fast an object is travelling, or in which direction, or how many times we have seen it previously, we must analyse several adjacent scans. Automatic Track Initiation (ATI) is therefore concerned with movement and history of an object. ATI serves as the final processing stage, the output of which is a track.

ATI is configured with the variables below:

  • Of the last 'N' scans, only create a track if we have seen it 'M' times (by default M=3 and N=5)

  • Only create a track from the plot if it’s travelling at a certain minimum and maximum speed (calculated as distance elapsed between scans divided by the scan time).

  • Only create a track if it is travelling in a straight line, +/- a certain error margin in degrees, called the turn rate (calculated as angle of direction change between scans).

In order to become a track, an object needs to exhibit various attributes and behaviour, e.g. physical size, speed, directional change rate etc. and in order to be classified, these attributes are measured against Classification Models. As such, classification takes place once an object reaches the state of becoming a track. This is ostensibly the point at which it appears in on the PPI, however in reality a track exists several seconds before appearing. 

For example,

Image Removed

Image Added

This image shows the process of 'ATIing' a plot. Using an M/N of 3/5, 3 plots (i.e. objects which satisfy the threshold and plot extraction criteria) lead to a track on the 4th plot onwards. Classification uses track data (as it needs speed/turn rate information), so in the example above, classification could not start until the 4th sample, unless provisional track information is used. Use of provisional tracks uses previous classification information available from previous plots which ultimately led to a track. This can result in a quicker classification. The downside (for a very low M/N) could be that the track is actually the result of a false alarm, so we end up with a classified false alarm. This should not be a problem if the tracker is tuned correctly.

Related information

Filter by label (Content by label)
showLabelsfalse
cqllabel in ( "witness" , "ati" , "classification" , "rule" , "rules" ) and space = "TUN"