Research Of Human Detection And Object Tracking Of Intelligent Video Surveillance

Intelligent video surveillance system is a popular field of computer vision research and new industries of commercial applications, widely used in the real-time monitoring of some security-sensitive place such as government agencies, banks, warehouses, residences, hospitals.Intelligent video surveillance mainly use the computer vision technology to process the camera's video stream, analysising and advanced understanding of the situation without the need for human intervention, through the analysis of image sequences on monitoring the changes of scene detection, identification and location tracking, then analysis the object behavior to predict, can provide useful information to effectively assist the security personnel when sudden crises occurs, reduce the manpower, material and financial resources and minimize false positive and false negative phenomenon. This paper base on the previous research, made some study and discussion in the field of intelligent video surveillance of human motion detection, segmentation and tracking.In moving target detection, the current major motion detection techniques are reviewed, focusing on the single Gaussian model, Gaussian mixture model and non-parameter model to estimate the main elements and ideas of the three algorithms of background modeling, then evaluate the performance by the experiment, and compare their advantages and disadvantages and effects.In moving target segmentation, introduce some common moving object segmentation algorithm:Based on the pixel gray value segmentation, region-based segmentation algorithm, based on the boundary of the segmentation algorithm, then divide the moving object by coordinate mapping division, which can make full use of the moving target detection results, with low complexity, can division the target accurately and quickly.In the human object tracking, present the algorithm of fast motion tracking algorithm combining Camshift algorithm and Kalman filter. First, discuss the basic principles of target tracking, then introduce some classical tracking algorithms and their advantages and disadvantages, focusing on the basic principles and application of MeanShift algorithm, and present the algorithm of fast motion tracking algorithm combining Camshift algorithm and Kalman filter, the experimental results prove the algorithm has good tracking results, and the tracking efficiency is desirable.