Research On Safety Monitoring Based On Human Posture For Night Vision

At present,the application of a rtificial intelligence is more and more widely,it has penetrated into various fields and industries,it is a trend that the machine will instead of man's labour.However,the vast majority of the monitoring systems that are being used on the market are no t intelligent,and they basically require human participation.After the accident,surveillance video often needs to be callback to understand the situation at that time.If you need to achieve the abnormal situation of the timely alarm,you need to monitor the scene all the time,which obviously can not meet the current needs,but also very prone to omission.Intelligent video surveillance system can monitor the scene to implement real-time monitoring,make the appropriate judgments according to the scene,and notify the staff to make the corresponding treatment to curb the occurrence of accidents in time and reduce losses without the participation of staff,through a number of video image processing algorithms.Human body attitude recognition is one of the core technologies of intelligent video surveillance technology,and it is also the most popular research direction recently.It includes many directions.This paper mainly studies two aspects of human posture,one is gait recognition in human posture,and the other is the fall detection in human posture.The main work of this paper is as follows:1.Research on Gait Recognition Algorithm Based on Angular Radial Transformation and Centroid.The gait recognition algorithm used in this paper is based on the combination of angle radial transformation and centroid.The algorithm is mainly aimed at the infrared image which has low contrast and lack of texture.The feature extraction module uses the overall appearance of the method,using the angle radial transform coefficient and the centroid vector as the eigenvector,and then through the method of PCA to reduce the dimension of the eigenvector,and the support vector machine to identify each individual.2.Research on Falling Detection Algorithm Based on Asp ect Ratio and Centroid Change Rate.This paper proposes an automatic fall detection algorithm combining human aspect ratio and centroid change.The algorithm mainly uses the feature of the human body aspect ratio to carry on the fall detection,and then use centroid change rate and effective area ratio to improve the accuracy rate.3.Development of Safety Monitoring System Based on Human Posture for N ight Vision.An automatic monitoring system is developed.The gait recognition algorithm and the automatic fall detection algorithm are introduced in the system,which can realize the intelligent monitoring of indoor al-weather.