Research Of Human Motion Pattern Recognition Based On Position And Attitude

With the increase of the aging population and the improvement of people's living standards,the health and quality of life of the elderly population has gradually received social attention.Falling down is one of the most dangerous accidents threatening the health of the elderly.The balance ability of the body has a great relationship with falling down.Hypertension,osteoporosis and other common diseases of the elderly also have a certain impact on the balance ability.At present,the research on the old people's body balance ability focuses on the rough evaluation of physical quality or the instantaneous change of the acceleration information of key parts of the body,so it is difficult to judge the balance ability accurately.Human motion pattern recognition is the basis of human intention judgment,health monitoring,human-computer interaction and other fields.At present,due to the complexity of equipment,it is difficult to popularize the picking features.Increase with the development of science and technology and the manufacturing process,give priority to with smart phones and smart watches smart devices constantly new people with intelligent wearable devices using viscosity increasing,using intelligent wearable device built-in low cost(low accuracy)motion sensor motion pattern recognition for single feature points into an irreversible trend.Smart devices are characterized by low cost,easy to wear,and easy to pick up and remove,and are characterized by frequent changes in the process of movement(including sitting and lying).The main research contents and related result are as follows.Based on the spatial position information of 42 key parts of the human body,this paper analyzed the 3-second walking process of 66 elderly people,and preliminarily extracted 25 characteristics used to measure the balance ability of the human body.Subsequently,the relationship between 25 features and the balance ability of the elderly was studied.The results showed that 13 features,including the gravity center and coronal plane deviation,the gravity center and sagittal plane deviation,and the degree of crotch,shoulder and arm deviation in the sagittal plane,could be used as indicators to measure the balance ability of the elderly.Finally,the 13 balance indicators were reduced to 3d and clustering analysis was carried out to distinguish the elderly with good balance ability and poor balance ability.The results showed that the balance indicators and clustering method after dimension reduction could clearly distinguish the two groups of people.Relationships between five common diseases and balance ability of 66 elderly people were studied,and the bayesian network model was adopted for modeling and analysis.The results showed that the probability of abnormal balance ability of elderly people suffering from hypoxia,cerebrovascular disease,osteoporosis and other diseases was significantly higher than that of normal elderly people.Develop a data collection APP,use the motion sensor built in android wearable device to collect posture information of human wrist,analyze the four processes of human walking,running,standing and descending,extract characteristic values such as signal mean,peak,variance and covariance,and adopt support vector machine algorithm for classification and recognition.Then,to improve the recognition accuracy,grid search method,genetic algorithm and particle swarm optimization algorithm were used to optimize parameters C and g respectively.The results show that this method can effectively identify the four movements of the experimenter:stationary,walking,running and descending.The research contents of this paperon can provide a reference basis for the measurement of body balance ability of the elderly,provide basic reference for judging human intention,health monitoring,human-computer interaction,and provide a research basis for further fall risk assessment.