
140
In addition, a better multi-feature fusion method could be designed to increase the classifica-
tion accuracy further.
4.3 Method comparison and result analysis
In general, as shown in Figure 4, the results of multi-feature fusion recognition were supe-
rior to single-feature recognition. The classification accuracy of feature-level fusion and
decision-level fusion increased 1.19% and 3.57% comparing to MAV, respectively; and
also increased 2.38% and 4.76% comparing to WL, respectively. From the point of each
subject, almost all the sequential movements were recognized by means of multi-feature
fusion, and the error of each subject was only ...