基于多传感器融合的用于脑卒中患者的可穿戴式上肢动作识别系统

目的:研究一种用于脑卒中患者上肢动作康复评估的可穿戴式的基于多传感器融合的上肢动作的自动识别系统。方法:提出一种基于多传感器融合的可穿戴上肢动作识别系统,以实现上肢动作的自动识别。系统硬件包含一个主节点和三个从节点,其中主节点由一个WIFI模块和一个微处理器构成,从节点由一个高灵敏度加速度计和一个微处理器构成。从节点采集上肢动作的加速度信号,传输至主节点,再由主节点的WIFI模块无线传输至PC端。这样的系统硬件具有功耗低、微型化、可拓展性强等特点。系统软件基于Adaboost的动作分类算法,通过分析接收到的上肢动作的加速度信号,自动判断不同的三种上肢动作。结果:系统对三种不同的上肢动作能够实现自动识别,识别准确率高达97.99%。结论:该可穿戴式系统在脑卒中患者上肢动作康复评估中具有良好的应用价值。

The wearable system based on the multi-sensor fusion for recognizing stroke patients' upper limb motion auto matically

Abstract Objective: To study a wearable system based on the multi-sensor fusion for recognizing stroke patients' upper limb motion automatically. Method: A wearable system was proposed based on multi-sensor fusion for recognizing stroke patients' upper limb limb motion automatically. The hardware of the system was made up of a master node and three salve nodes. The master node consists of a WIFI module and a microprocessor, and the slave node was made up with a highly sensitive three-axis accelerometer and a microprocessor. Accelerometer signals were collected by the slave node, then transmitted to the master node, after that, transmitted to the computer through the WIFI. This hardware system had characteristics of the low power consumption, miniaturization, expansibility and so on. The algorithm for motion classification based on Adaboost was realized in the software of this system, which would recognize three different motions automatically through analyzing the received acceleration signals of upper limb motion. Result: Three different motion of upper limb could be recognized automatically by the system, the accuracy of motion recognition was up to 97.99%. Conclusion: The wearable system has great value in rehabilitation assessment of upper limb motion for stroke patients.