Controls and Control Theory Commons^™

Statistical Movement Classification Based On Hilbert-Huang Transform, Shanqi Sun

Graduate Theses and Dissertations

The goal of this project is to introduce an automatic movement classification technique of

finger movement signals using Hilbert-Huang Transform (HHT). Due to the nonlinear and

nonstationary processing behavior, movement signals are analyzed with the Hilbert-Huang

Transform (HHT). The slope of auto-correlation function and mean of frequency from first

three Intrinsic Mode Functions (IMFs) was used as feature parameters for each category.

Finally, performing support vector machine (SVM) for pattern classification completes clas-

sifying types of finger movement. According to the records of 669 trial samples of two types

of finger movement signals (thumb and pinky), average accuracy is 93.28%. …