Phantom: Prediction of Human Motion with Distributed Body Sensors - Prediction and Classification of Respiratory Motion - page 56

methods. That means head motion can be classified into different groups in the
given datasets. For our tracking estimation, we use grouping number G = 3 for
Head datasets because of the minimum AIC value in the given results.

3.5.4 Multi-Channel IMME

Based on the group number (G*) chosen in the experiment Sect. 3.5.2 of the first
stage, we can perform the target estimation using Multi-channel (MC) IMME of
the second stage with respect to each group.

(1) Position Estimation

We compare the performance of motion tracking estimation among KF, IMME,
and MC-IMME. Figure 3.12 shows that MC-IMME can estimate the target motion
more accurately than other tracking methods, KF and IMME at the initial stage.
In addition, we compare the accumulated position errors for each channel across
the entire measurement period among KF, IMME, and MC-IMME. Figure 3.13

Fig. 3.12 Comparison of
motion tracking estimation
for Head_1 dataset

14.8

Measurement
Kalman Filter
IMME
MC-IMME

14.75

14.7

14.65

14.6

14.55

14.5

0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1

Data Time Index (second)

Fig. 3.13 Comparison of
accumulated position error of
each channel for Head_1

18

Kalman Filter
IMME
MC-IMME

16

14

12

10

8

6

4

2

0

1 2 3 4 5 6 7 8

Sensor Numbe