Research of elderly fall detection based on Dynamic Time Warping algorithm
Huan Li, Yongli Yang · 2016
With the acceleration of the aging process of human society, home health monitoring systems have drawn great attentions of the researchers. In this paper, a novel fall detection algorithm based on DTW (Dynamic Time Warping) method was proposed and a fall detection system which could be implemented on wearable devices was designed. In the algorithm, the body motion data acquired by acceleration sensor and gyroscope is firstly preprocessed and fused by quaternion coordinate correction and Kalman filtering, then the LPCC (Linear Prediction Cepstrum Coefficients) features of the body motion data were analyzed and extracted. Finally, the extracted features were matched to pre-acquired fallen templates by DTW algorithm. Experiments confirm that our algorithm and system could precisely and effectively recognize the fall state of the human body.