The Witness software will attempt to reliably classify all targets based on all the attributes available from the radar sensor. It does this using a weighted scoring system based on a series of pre-defined models. Each model provides a series of value ranges and the prioritises the values which are more relevant to that classification type. When a target under analysis conforms to the specified values a weighted scoring is added for the matching model or models. Once the agreed number of classification samples have been processed then the model which has achieved the highest score is selected. A probability is assigned to each classification based on the score where 100% is a classification with the highest confidence.

Since V3.12 the classification has been improved to work with aggregated values, which means we can sample the data, apply a suitable statistical summary (i.e. min, max, average etc) and then use these values to test each model. This has proved to be a more reliable way of distinguishing between targets which are exhibiting similar behaviour to another classification type (i.e. a vehicle vs a person when the vehicle is moving at walking pace).