This paper discusses preprocessing in a tracking filter to estimate th
e parameters of target motion, such as position and velocity. We use o
rthogonal coordinates, with the target position as the radar observati
on data. Typical examples of tracking filters are the Kalman filter an
d the alpha-beta filter. In a Kalman filter, although the tracking acc
uracy is high, the computation load is heavy. Consequently, a simplifi
ed version of the Kalman filter, the alpha-beta filter, is used. The a
lpha-beta filter has problems in tracking accuracy, although the compu
tation load is low. Another difficulty with the alpha-beta filter conc
erns its application to radar tracking when the observation noise vari
es with sampling time or when the sampling interval is non-uniform. In
such cases, the Kalman filter must be used to achieve adequate tracki
ng performance. This paper proposes a method in which the tracking fil
ter consists of a Kalman filter with preliminary integration of multip
le observation data, obtained at different sampling times, into a sing
le observed datum. It is shown that the proposed tracking filter can b
e approximated by a tracking filter in which a Kalman filter is used f
or each sample. The tracking accuracy and the computation time are eva
luated by computer simulation. (C) 1998 Scripta Technica.