A human fall recognition method based on acceleration sensor

Abstract

The present invention belongs to the human body action identification technology field, in particular relates to an acceleration sensor-based human body falling down identification method. Aiming at the problem that a current human body motion identification technology neglects the performance and the power consumption of an intelligent mobile phone, the present invention provides the accelerationsensor-based human body falling down identification method. By designing a feature extraction method irrelevant with the mobile phone carrying position, and based on the features, a decision tree-based falling down identification method is provided. An identification algorithm is designed for an intelligent mobile phone platform, fully considers the performance, power consumption and real-time performance requirements of the intelligent mobile phone, has the characteristics of being small in operand and little in resource consumption and being able to operate for a long time, at the same time, has the higher accuracy. In addition, the method fully exploits the portable advantage of the intelligent mobile phone to the full, provides the all-weather human body motion identification servicefor the users, can timely discover the falling down behaviors of the users and sends out the alarm information, so that the users can be helped at first time, and the falling down harm is reduced to the lowest.

Description

technical field [0001] The invention belongs to the technical field of human body action recognition, and in particular relates to a human body fall recognition method based on an acceleration sensor. Background technique [0002] In recent years, the proportion of the elderly population in my country has shown an increasing trend year by year. With the flow of the working population, the phenomenon of "empty nest elderly" has become more and more. The support and care of the elderly has gradually become the focus of society. Among the problems, falls are the most common problem for the elderly. Falls are extremely harmful to the elderly. If they are not rescued in time, they may even be life-threatening. With the intelligentization of mobile devices, mobile smart phone platforms are developing at an unprecedented speed, and it has become an industry consensus that smart phones integrate a large number of sensors. Among them, the acceleration sensor has become a standard acc...

AI Technical Summary

Problems solved by technology

However, visual recognition technology cannot be accurately recognized when the light is poor, and it is easy to leak user privacy. However, there are very few products for human motion recognition based on acceleration sensors. At present, most human motion recognition technologies based on acceleration sensors have a large amount of calculation Large, poor portability, researchers pay little attention to performance bottlenecks and power consumption bottlenecks of mobile phone platforms

Since fall detection requires uninterrupted operation, and it is difficult for smart phones to perform large-scale data calculations for a long time, the present invention mainly studies the following problems: 1. The position of the mobile phone is not fixed, resulting in the existence of collected data. The big difference is not conducive to the accurate recognition of human body movements; second, limited by the hardware performance of the smartphone platform, the current smartphone CPU and memory are limited, and large-scale data processing cannot be performed, and the hardware configuration of the mobile phone platform is uneven. Hardware based on ARM, MIPS, X86 and other architectures is difficult to guarantee the compatibility of program transplantation; 3. Recognition of human body motion requires a certain real-time performance, or a very small delay is allowed, which means that it must be implemented locally on the client side. Recognition algorithm; 4. Human motion recognition needs to run continuously for a long time without interruption to avoid memory leaks, power consumption, long program running time, mobile phone heating, freezing, etc.

Images