Research on Recognition of Nine Kinds of Fine Gestures Based on Adaptive AdaBoost Algorithm and Multi-Feature Combination

• Published in: IEEE access Vol. 7; pp. 3235 - 3246
• Main Authors: Liang, Shili, Wang, Liangliang, Zhang, Ling, Wu, Yansheng
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 2019; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: AdaBoost algorithm; Adaptive algorithms; Algorithms; Classification; Decision trees; Feature recognition; Gesture recognition; multi-feature combination; Myoelectricity; Prostheses; Segments; surface EMG signals
• ISSN: 2169-3536; 2169-3536
• DOI: 10.1109/ACCESS.2018.2884773

Accurate recognition of gestures based on surface EMG signals is of very importance in the study of human prosthetic interaction. In this paper, multi-feature combination and adaptive AdaBoost algorithm are used to identify nine kinds of fine gestures. First, in order to adaptively set the threshold for extracting the active segment, the algorithm for detecting the active segment of the surface myoelectric signal is improved, and the effective active segment value is obtained by rejecting the invalid active segment. Second, a fine gesture classification method based on multi-feature combination is proposed. In this paper, the recursive feature elimination algorithm based on cyclic feature selection is used to screen out the feature set with the top classification effect, and the feature set selected by eight experimental objects is counted, and the optimal feature combination is obtained according to the experimental analysis. Finally, nine fine gesture actions are identified through a weak classifier constructed by combining the AdaBoost algorithm and the single-layer decision tree. The experimental results show that the proposed algorithm can achieve an average recognition rate of 98.56 (±1.44)% for nine fine gestures.