Extended Target Tracking using Principal Components

• The increased resolution in today’s radar systemsenables tracking of small targets. However, tracking both smalland large targets in a dense target scenario raises considerablechallenges. The data association of tracks to measurement groupsis highly dependent on good target extension models for filteringand likelihood computation. In our attempt to design a trackerfor extended targets, we start by adopting the results from thetechnique referred to as random matrices, which enables us toseparate the filtering into an extension and a kinematical part.We re-define the measurement model and discard the assumptionof independent Gaussian-distributed plots. Instead we assume theprincipal components to be Gaussian distributed. Then, througha heuristic approach, we create a two-stage Kalman filter, wherethe first stage estimates the principal components, and the secondstage estimates the centre of gravity, using the output fromthe first stage as measurement uncertainty. The advantage ofhaving a Kalman filter with data-driven measurement noise overa standard Kalman filter is demonstrated using simulated data,where a significant improvement in terms of smaller errors andreduced track loss is shown.

Ämnesord

TEKNIK OCH TEKNOLOGIER  -- Elektroteknik och elektronik -- Signalbehandling (hsv//swe)

ENGINEERING AND TECHNOLOGY  -- Electrical Engineering, Electronic Engineering, Information Engineering -- Signal Processing (hsv//eng)

TEKNIK OCH TEKNOLOGIER  -- Elektroteknik och elektronik -- Annan elektroteknik och elektronik (hsv//swe)

ENGINEERING AND TECHNOLOGY  -- Electrical Engineering, Electronic Engineering, Information Engineering -- Other Electrical Engineering, Electronic Engineering, Information Engineering (hsv//eng)

Nyckelord

random matrices

extended targets

Kalman filtering

Target tracking

principal components

Publikations- och innehållstyp

kon (ämneskategori)

ref (ämneskategori)