| The intelligent monitoring system is basically a system of surveillance and recording by accessing video image information.This also helps in making safety precautions.This is widely used in milit ary,customs,public security,fire protection,forestry,railways and other public places.Intrusion detection technology based on infrared video image has become a hot topic in the field of intelligent surveillance.In this paper,we will study combinati ons of the embedded processing platform and pattern recognition technology,to realize indoor intrusion infrared intelligent monitoring system and alarm.This paper introduces the development of video surveillance technology,the latest research of the pedestrian detection algorithm and embedded image-processing system.Firstly,the foreground objects in the infrared video image are extracted.Compared with the background difference method,the MSER and Vibe algorithms are used to extract the foreground mo ving objects,and the Vibe algorithm has a better segmentation effect.Secondly,the paper describes three typical feature extraction algorithms including Invariant Moment?DCT and HOG to realize contour extraction.Through comparison,the HOG method is used to describe the characteristics of infrared human.Based on the diversity of human posture and indoor intrusion detection,this paper makes the human database under infrared video,including upright,bending,fat and thin,different distance and so on.In addition,using AdaBoost to realize the classification of human objection in infrared surveillance video.Finally,the proposed algorithm is transplanted to the embedded development platform to achieve the identification of indoor intrusion targets and generate alarm signals.Through the experimental verification and testing,the detection algorithm proposed in this paper can effectively identify indoor intrusion objects.The overall recognition rate is as high as 88.36%.The experimental results show that it has certain engineering application value. |
PDF Full Text Request