Robust Motion Recognition Techniques Using Orientation Sensor Signals

• Main Authors: Sheng-Jie Lin, 林聖傑
• Other Authors: Ching-Hsiang Tseng
• Format: Others
• Language: zh-TW
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/09403411196862401902

碩士 === 國立臺灣海洋大學 === 電機工程學系 === 98 === The objective of this research is to develop advanced techniques for enhancing the humanization of man-machine interfaces. Specifically, we intend to create a motion recognition system which can properly recognize motions with arbitrary orientations usigan orientation sensor. Research work on the recognition of tennis strokes using acceleration signals can be found in the literature. Note that the strong movements and distinct features of the tennis stroke motions make them relatively easy to be recognized. In this paper, we consider the recognition of the writing motions for the Arabic numbers 0 to 9. Note that these are localized motions with significant similarities, which make them relatively difficult to be recognized. To solve this problem, we utilize the hidden Markov model as a basis to construct a user independent motion recognition system which can accommodate motions with arbitrary orientations. This means that the system allows the user to write the Arabic numbers in an arbitrary orientation. The system analyzes the acceleration and Euler angle signals of the writing motions, performs signal processing and feature extraction, and yields the recognition results. The experimental results indicate that a recognition rate of 90\% can be achieved for a user-independent task. This justifies the robustness of the proposed method for recognizing writing motions under arbitrary orientations.