Old people stumbling detection method based on handset

Abstract

The invention discloses an old people stumbling detection method based on a handset, comprising: collecting the information of the triaxial accelerometer and the gyroscope sensor of a handset to obtain an acceleration and an angular acceleration; employing a wavelet filtering method to filter the collected acceleration and angular acceleration to obtain a post-filter acceleration and angular acceleration; constructing time domain characteristic vectors of the post-filter acceleration and angular acceleration, and employing an improved SVM (Support Vector Machine) learning classification algorithm to perform a first level filtering on the characteristic vector of the stumbling acceleration of the triaxial accelerometer; obtaining a second level stumbling determination result through the characteristic vector of the stumbling angular acceleration of the gyroscope sensor; and finally determining whether a person stumbles and completing stumbling identification. The old people stumbling detection method is used for detecting stumbling, and has the characteristics of high accuracy, short detection time, and low cost; the method can be applied only through taking a handset, and is convenient to use.

Description

technical field [0001] The invention relates to the technical field of electronic information detection, in particular to a mobile phone-based elderly fall detection method. Background technique [0002] The 21st century is known as the "silver-haired century". In the world, the wave of aging has already rolled in. According to the statistics of the Ministry of Civil Affairs in 2013 and the "China Aging Development Report" issued by the National Office on Aging, by the end of 2012, the number of elderly people aged 60 and above had reached 194 million, accounting for 14.3% of the total population. According to the "China Injury Prevention Report" released by the Ministry of Health, falls are the primary cause of accidental injuries for the elderly. Among the community elderly over 65 years old in my country, at least one-fifth of men have fallen, and nearly one-half of women have fallen. Falls are one of the important causes of disability, disability and death in the elderl...

AI Technical Summary

Problems solved by technology

However, most of the existing fall solutions are only applied to external wearable devices. There are few fall detections based on the mobile phone platform, and most of them only use acceleration sensors. The collected sensor information containing a lot of noise is not filtered. The algorithm is relatively single, most of which use threshold judgment algorithms, the detection accuracy and real-time performance are not high, and a few schemes that use machine learning algorithms require a long learning time and are not suitable for mobile devices

Images