Research on Elderly Fall Prevention Recognition System for Intelligent Elderly Care and Machine Vision

In order to improve the sensitivity and specificity of elderly fall recognition, a set of elderly fall prevention recognition system based on intelligent elderly care and machine vision is designed. The system was jointly developed based on the EAIDK-310 embedded development board and OpenCV library. The moving object detection algorithm based on the hybrid Gaussian model was adopted to detect the moving objects, and the multi-parameter comprehensive evaluation algorithm based on the three contour features of human aspect ratio, inclination angle and centroid height was adopted to recognize the fall behavior, thereby realizing the elderly fall recognition. The test results show that the average sensitivity and specificity of this system for the elderly fall behavior recognition is 94.23% and 95.86%, and the system can process about 10 frames of images per second, which has a certain realtime performance and can realize realtime recognition of the elderly fall behavior.