一种基于聚类分析的自适应步态检测方法

An Adaptive Gait Detection Method Based on Clustering Analysis

提出一种基于K-中心点聚类算法的自适应步态检测方法,检测不同步态参数及其耦合关系.所提方法在现有检测方法的基础上增加了步态精细划分环节,提高步态检测结果的正确性和有效性.实验结果显示,在较大步态参数空间内,采用所提检测方法可将步数估计的精度从现有方法的46.16%~53.22%提高到76.13%.

Gait analysis was one of the most focusd research fields in recent several years, and the gait param- eters attracted increasing interest in clinical medicine, pedestrian navigation and so on. However, the existing gait detection methods had some shortcomings that prevented their successful use to many practical applica- tions, the detection results of which were very sensitive to measurement fluctuations and detection parameters, and thereby characterized by poor robustness. In this paper, the mutual coupling relationship between different parameters was tested, and an adaptive gait detection method based on clustering analysis was proposed, so as to automatically yield the time heuristic threshold. The experimental results demonstrated the correctness and effectiveness of the method, and the gait detection accuracy over a large parameter space could be improved from 46.16% and 53.22% respectively to 76.13%.