Design And Implementation Of Indoor Monitoring System For Elderly People Living Alone Based On Kinect

The rapid development of the social economy has brought about many problems such as the aging of the population and the rapid increase in the number of empty nesters.People are paying more and more attention to the safety of the elderly.Therefore,it has great significance and development prospects to intelligently detect and analyze the behavior of indoor elderly.After analyzing the daily behaviors of elderly people,this paper designs and implements a kinect-based behavior monitoring system to monitor the daily behavior of the elderly and provide a safer living environment for the elderly living alone.The main function of the system is to monitor the daily behavior of the elderly and identify normal and abnormal behaviors.The normal behavior is divided into five categories:walking,standing,sitting,lying,squatting.And taking into account the time factor,if the sitting and lying behavior lasts longer than the specified time,it is defined as a suspected coma,and can be classified as abnormal behavior with falling behavior.When the system detects a fall and suspected coma,it immediately issues a warning to the guardian.For fall behavior,the system generates a record of the fall event,including time,location,how to fall,and so on.The behavior recognition process in this paper is realized through comprehensive judgment of human body movement posture,human body position information and 3D environmental information.Posture analysis is to extract the human body into different postures,and divide the human body into three postures of upright,curved and lying.The position information of the human body is obtained by detecting the coordinates of the human body in the two-dimensional image through human body target detection.And using kinect can obtain depth information to calculate the depth distance histogram,realize the distance-based segmentation,and obtain the distance range from the human body to the camera;3D environment information refers to the object coordinate information obtained by the indoor object detection and the distance range of each object to the camera,using these information realizing 3D environment reconstruction.In the comprehensive judgment,the relationship between the person and the object can be judged by analyzing the coordinates and distance information of the person and the object,then combined with the posture classification result to realize the behavior classification.At the same time,combined with the time factor,the duration of each type of behavior is monitored to identify the suspected coma;combined with the historical state of motion,the fall behavior is analyzed in more detail,and the falling state is obtained,such as falling when walking,falling when getting out of bed,and the like.The experimental results show that the system can monitor the daily behavior of the elderly in real time,identify normal and abnormal behaviors,and have better fall recognition for different fall postures.