Abstract—The most common moving processes include walking, running, going up stairs, and going down stairs. People distinguish the first two types of movement based on their perception of speed. However, people’s values are different, making it difficult to distinguish walking from running objectively. On the other hand, speeds of walking are usually not far from speeds of going upstairs/downstairs. Thus, it is difficult to train a computer to objectively distinguish these four types of movement only based on “speed”. This study tried to find a way to quantify movements during the experiments. Also a sensor was used in the experiments to help obtain the data during the movements. The sensor could continuously record the detailed information of each movement using packets. The pattern of each movement was represented using the corresponding acceleration data along the three axes of this study, the x axis (horizontally moving back and forth), the y axis (vertically moving up and down), and the z axis (horizontally moving in and out). Through the Fast Fourier Transform, the differences among various types of movement were observed. The extracted information was further used as the basis for computers to distinguish the four types of movement, walking, running, going upstairs, and going downstairs.
Index Terms—Fast fourier transform, walking-pattern, running-pattern, motion profile.
C. H. Lee and J. H. Liou are with Chaoyang University of Technology, Taiwan R.O.C. (e-mail: chlee@cyut.edu.tw).
[PDF]
Cite: Chu-Hui Lee and Jyun-Han Liou, " Movement-Type Classification Using Acceleration Sensor," International Journal of Information and Electronics Engineering vol. 4, no. 6, pp. 490-494, 2014.
