A Fall Posture Classification and Recognition Method Based on Wavelet Packet Transform and Support Vector Machine

Qingyun Zhang,Xianyi Zeng,Qinglin Sun,Matthias Dehmer,Jin Tao,Quan Zhou

doi:10.3390/app11115030

Qingyun Zhang, Xianyi Zeng + Show 4 more

Open Access

https://doi.org/10.3390/app11115030

Copy DOI

Abstract

An accidental fall seriously threatens the health and safety of the elderly. The injuries caused by a fall have a lot to do with the different postures during the fall. Therefore, recognizing the posture of falling is essential for the rescue and care of the elderly. In this paper, a novel method was proposed to improve the classification and recognition accuracy of fall postures. Firstly, the wavelet packet transform was used to extract multiple features from sample data. Secondly, random forest was used to evaluate the importance of the extracted features and obtain effective features through screening. Finally, the support vector machine classifier based on the linear kernel function was used to realize the falling posture recognition. The experiment results on “Simulated Falls and Daily Living Activities Data Set” show that the proposed method can distinguish different types of fall postures and achieve 99% classification accuracy.

Highlights

This paper proposes a fall posture classification and recognition method based on time series analysis and uses support vector machine (SVM) to distinguish between various fall postures and daily behaviors, achieving 99% classification accuracy, and good real-time performance; wavelet packet transform (WPT) was used to extract multiple features from sample data, and random forest was used to evaluate the importance of the extracted features and obtain effective features through screening
It was pointed out that [33] there is no common standard for evaluating the effect of fall posture recognition at this stage, and it is challenging to perform effect testing based on general procedures
The results indicate that SVM has significant advantages in solving the problem of small samples and highdimensional features, and it confirms the effectiveness of the wavelet packet transform in feature extraction

Summary

Introduction

Once an elderly person falls, the degree of injury of the elderly may be more serious if no one else finds or rescues them, the old person’s injury may be more serious, and they may even fall into a coma, endangering their life. Different fall postures during a fall will cause different impact positions, causing varying degrees of injury to the elderly [4,5]. There is a need for an automatic detection method, which cannot only recognize the fall behavior of the elderly in time, and recognize the fall posture, so that the nursing staff can grasp the specific details such as the impact position and the degree of injury of the elderly, and provide the elderly with more targeted rescue and treatment

Methods

Results

Discussion

Conclusion