應用在人機互動當中基於慣性感測器的一種新的手勢辨識技術

為使人機互動能更加地直覺便利，不少研究藉由慣性感測器，開發出手勢辨識系統來讓人和機器互動。在這篇論文當中，我們分析Wii Remote和 Wii MotionPlus當中的加速規和陀螺儀所感測到的加速度和角速度，並利用這些訊號辨識出使用者所表現的手勢。不同於以往直接辨識完整的訊號，我們首先根據手勢呈現的角度，將完整的手勢訊號拆解成不同的子手勢。除此之外，我們提出了一種改良式的k-means分群演算法，利用差異演化演算法 (Differential Evolutionary Algorithm) 來找出最佳的分群初始點，此改良式的k-means分群演算法被用來量化每個子手勢以降低手勢當中的變異性。最後，我們利用兩層的階層式隱馬可夫模型 (hierarchical hidden Markov model) 來辨識使用者所表現的手勢。其中，第一層的隱馬可夫模型被用來辨識一個完整手勢當中所包含的子手勢，而辨識過的結果會進一步被重組，當成第二層隱馬可夫模型的輸入，做第二階段的辨識。為了驗證我們所提出的方法，我們設計出一個包含12種不同手勢的手勢庫來做測試，實驗結果顯示出我們提出的手勢辨識技術可以在平均81 ms的時間內成功地辨識出使用者所表現的手勢，其正確性高達94%。

關鍵字

手勢辨識；差異演化演算法；隱馬可夫模型；加速規；陀螺儀

並列摘要

To make the human computer interaction (HCI) more intuitive and convenient, numerous works have been applying inertial sensor-based gesture recognition to remotely interact with the computer. In this thesis, we analyze the data sensed by the accelerometer and gyroscope in the Wii Remote which is attached by the Wii MotionPlus to recognize the currently performed human gesture. Instead of directly analyzing the raw data sequences, we take a different approach to decompose the complete gesture trajectory into several sub-gesture trajectories according to the “orientation angle” of the gesture. Furthermore, a modified k-means clustering algorithm which uses the differential evolutionary algorithm to select the optimal initial points is used to model each sub-gesture in order to cope with the intra- and inter-variations of the gestures. Finally, a two-layer hierarchical hidden Markov model (HHMM) is adopted to recognize the gesture where the lower layer HMM is used to recognize the sub-gestures, and the results of the lower layer HMM are assembled and input to the upper layer HMM for the second stage recognition. To validate the proposed approach, we here design a gesture set which consists of ten different gestures and is more than most of the other datasets. The experimental results show that the recognition process of our system takes about 81 ms and the recognition rate is higher than 94%.

並列關鍵字

Gesture Recognition ； Differential Evolutionary Algorithm ； Hidden Markov Model ； Accelerometer ； Gyroscope

參考文獻

[4] Jiayang Liu, Zhen Wang, Lin Zhong, Jehan Wickramasuriya, Venu Vasudevan, "uWave: Accelerometer-based personalized gesture recognition and its applications," IEEE International Conference on Pervasive Computing and Communications, pp. 1-9, 2009.

[1] S. Goldin-Meadow, "The role of gesture in communication and thinking", Trends in Cognitive Sciences, vol. 3, pp. 419-429, 1999.

[2] Jeen-Shing Wang, Yu-Liang Hsu, Jiun-Nan Liu, "An Inertial-Measurement-Unit-Based Pen With a Trajectory Reconstruction Algorithm and Its Applications," IEEE Transactions on Industrial Electronics, vol.57, no.10, pp.3508-3521, Oct. 2010.

[3] H.-I Suk, Bong-Kee Sin, Seong-Whan Lee, "Robust modeling and recognition of hand gestures with dynamic Bayesian network," 19th International Conference on Pattern Recognition, (ICPR’08), pp.1-4, 2008.

[6] Chun Zhu and Weihua Sheng, “Online Hand Gesture Recognition Using Neural Network Based Segmentation”, IEEE/RSJ International Conference on Intelligent Robots and Systems. October 2009, pp. 2415 - 2420.

國際替代計量

應用在人機互動當中基於慣性感測器的一種新的手勢辨識技術

全文下載

主題瀏覽