The Design Of Intelligent Video Monitoring System For Elderly Apartment

The aging population has become the inevitable trend of social development,and the elderly care for the elderly is widely concerned by the whole society.With constantly improve people's living standard in China,the traditional family pension mode is facing many challenges: family size decreases,the lack of care resources,the demand of the elderly is becoming more and more rich,these factors led to a new type of pension mode-the elderly apartment pension arises at the historic moment.With the increase of the elderly apartment,the safety of the elderly has become a public concern.In China,the number of elderly apartment service personnel is insufficient,and frequent falls of elderly people cannot be found in time,leading to major accidents.The video monitoring system with attitude recognition technology can solve this problem.Will be based on the image of human body gesture recognition technology is applied to the elderly apartment of the video monitoring system,by means of monitoring the processing of the image,to extract the target information,and according to some characteristics of the human body target,to analyze and identify the body posture,gesture automatic judgment to human body,if it is found that the old man fell down the abnormal phenomenon such as timely report to the police.In this paper,human body target detection,feature extraction and attitude identification technology are studied in the fixed scene.Firstly,based on the introduction and comparative analysis of several main target detection technologies,this paper puts forward the algorithm of classification background model estimation used in this paper,which is very good for human body target detection.Then the human body target is extracted.The edge detection method is used to extract the edge characteristics of human body.The second step is to extract the eigenvectors of the human body on the basis of segmentation.Finally,according to the sample image to extract the feature vector,using the BP neural network,according to the stand,bend,down three kind of body posture to the BP neural network training,to establish BP network model,using BP neural network model to identify the body posture.