TARGET TRACKING WITH INFRARED IMAGING AND
MILLIMETRE-WAVE RADAR SENSOR
ZHANG Xuejing
1
2
,MA Long
1
,CHEN He
1,
*, YANG Jing
2
(1. Departmen
of Electronic Engineering, School of Information
and Electron, Beijing Institute of Technology, Beijing 100081,
China; 2. Department of Electronics Engineering
School of Information, Beijing Union University
Beijing, 100101
)
Corresponding author(email:chenhe@bit.edu.cn)
Abstract
Two commonly used tracking fusion methods for Kalman-
filter-based multi-sensor data fusion which are weighed cross-
covariance fusion and augmented measurement fusion are
analysed in this paper. Based on tracking fusion of infrared
sensor and millimetre wave Radar ,the fused states and
measurements are compared with individual estimates.
Results are presented using Monte Carlo simulation by two
given virtual trajectories which show that: (1)the two fusion
methods are functionally equivalent if the sensors used for
data fusion have identical measurement matrix;(2)the
obtained joint state-vector estimate is better than the
individual sensor-based estimate. Also presented are the
possible reason caused the bias between individual position
estimate and true followed by the analysis of the
computational advantages of each method.
Keywords: Extended-Kalman filter; infrared Image;
millimetre wave Radar ˗ Weighed cross-covariance fusion;
Extended measurement
1 Introduction
With the rapid development of sensor technology, a variety
of multi-sensor system[1-7] aiming at different complex
applications background spring up at an increasing rate.
Because the battlefield surroundings are becoming more and
more complicated, the seeker with only one guiding sensor
can easily be cheated and interfered. In other words, it can
hardly meet the demand of precision target recognizing and
tracking. Therefore multiple sensors are widely used to
enhance target-tracking capabilities[8,9]. Radar can measure
range with good resolution, but angular measurements have
poor resolution. An infrared search-and-track sensor (IR) can
measure the azimuth and elevation of a target with good
resolution. It can provide only the direction of a target but not
its location because it does not provide the range. The fusion
of measurements from radar and IR results in less uncertainty
of the estimated position of the target.
Figure.1 A Infrared/Millimetre Wave seeker Tracking
schematic diagram
Figure 1 shows the schematic diagram of a millimetre wave
radar/ infrared composite guidance systems .Infrared sensors
and millimetre-wave radar are used to detect the tracking
target ; Two sets of servo provide servo drive for infrared and
radar sensors; The trackers are used to achieve the automatic
recognition and tracking of the target; The controllers are
used to coordinate the work of sensor, servo , tracker and
communications for information fusion ; information fusion is
responsible for the communication to Navigation , and to
coordinate the work between infrared and radar .The guided
strategy is mainly generated in the information fusion .
Two methodologies[10] have been employed for track fusion
purpose, which are state vector fusion and measurement
fusion. there are two approaches to state vector fusion, which
are :weighted cross covariance and information matrix. But
the performance of later is influenced greatly by the
feedback(particularly the fused covariance) which is not exact
and could mislead the local sensors. There are also two
measurement fusion methods for multi-sensor data fusion.
The first combine the multi-sensor data based on a minimum-
mean-square-error criterion but may be inapplicable in some
situations, such as with dissimilar sensors whose
measurement matrix are of different size[11].The second
called augmented measurement fusion merges the multi-
sensor data through the observation vector of the Kalman
filter and implement is simple. This research is restricted to
the weighted cross covariance fusion and augmented
measurement fusion.
The two fusion methods analysed in this paper are
concentrated approach and distributed approach respectively.
In the first method, the measurements from IR and radar are
combined using covariance matrix as weights; and in the
Millimeter
Wave Radar
Infrared Imaging
Sensor
Radar Servo
Radar Track
Infrared Servo
Infrared Track
Radar
Control
Infrared
Control
Data
Fusion
Navigation