Research And Implementation Of Behavior Of Identification Technology In City Security Monitor System

The efficient algorithms of body (group) behavior detection and analysis under Complexbackground is the forefront of research topics in the field of computer vision, patternrecognition, which is a key technology for intelligent monitoring system, has a very broadapplication prospects in the field of national security, civil. In real life, a large number ofmeaningful visual information contained in the motion, the key of image processing for futureanalysis is how to extract the moving target from the real-time changes in the backgroundquickly and accurately detect then out. Moving target detection has the important scientificsignificance and application value in gesture analysis, human behavior recognition, and theemergencies groups. In order to complete the higher level the visual task, Online detectiontarget tracking, precise positioning, get the target’s motion parameters for subsequentprocessing and analysis, and understanding the behavior of the moving target are necessary toperform.In this paper, firstly, we briefly describe the complex group behavior in recognition of theurgency and necessity, and secondly, we generally discuss the status quo of domestic andinternational technology group behavior recognition under complex background. Thirdly, wehighlight the behavior analysis of previous work, namely the detection of moving targets. Aswe all know, the effect of the moving target detection has an important effect on thesubsequent behavior analysis, detailed moving target detection method. So we focus on thegeneral steps of moving target detection and some famous moving target detection algorithms.In this paper, we carried out a detailed analysis of the moving target detection from the aspectof fast update algorithm and slow update algorithm, then combine the two algorithms. In theactual use of combining the two methods to achieve the perfect the effects still need thesubsequent processing including the expansion and corrosion, a series of processing work toeliminate glitches and fill voids, as well as vanishing.Next, we highlight the behavior recognition technology. In this area, we mainly introducethe feature extraction method which is based for behavior recognition methods, such assolid-state characteristics, dynamic characteristics, the spatial and temporal characteristics,and the behavior of the final common behavioral modeling technology, such as templatematching model, probabilistic graphical models, grammar-based model, and the model basedon statistical relational learning; then we propose the behavior recognition algorithm based on latent semantic analysis PLSA. In the fourth chapter, we highlight the target tracking,introducing the popular target tracking algorithm in detail, and analyzes their advantages anddisadvantages, and then discuss the technical difficulties and key points as well as the ways tocope, and the specific details of the experimental demonstration and presentation from thesingle-target tracking and multi-target tracking two aspects in detail. the tracking results ofeither single or multi-target tracking target tracking are very good. Also simple narrativetarget behavior focuses on the two key goals of behavior from the classification of the targetbehavior, aggregation and single target separation of the two goals. Behavior for the two goalsare simply introduced and analyzed.This paper is based on behavior recognition technology of urban security monitoringsystem, which requires complex context groups behavior analysis and research (group) of thereal-time detection of moving targets, tracking and behavior recognition and understanding,study the visual information to calculate the cognitive mechanism to build a newcomputational model and efficient calculation methods to improve the understanding of visualperception ability and processing efficiency. The target detection in the video monitoringobject is from the sequence of images in the region of interest from the background imageextracted. However, due to the dynamic changes of the background image, such as the impactof weather, light, shadow, occlusion and disruptors, surveillance cameras captured the trafficon the road, in the sequence of images may include pedestrians, vehicles, and others, such asbirds, clouds and shaking branches and other moving objects, the target detection and trackingto be a very difficult job. Currently, research on body behavior recognition is mostly limitedto the strict experimental environment, the development of robust, high stability of humanbehavior recognition system and its application, and also need a lot of research work.