Reserch Of Video Moving Object Detection Algorithm Based On Gaussian Mixture Model And Its Application

Video-based moving object detection has been widely used in video surveillance,intelligent transportation,video encoding,human-computer interaction.Currently,scholars made a lot of target detection algorithm,these detection algorithm have achieved very good results in the hand of stability and accuracy,but there are still difficult to adapt to environmental changes influence and hard to meet real-time requirements of the problem.Therefore,the study of the moving target detection technology is still a challenging task.In the field of intelligent traffic monitoring,traditional video surveillance system is to achieve a basic video capture,storage,video playback and other functions,video just rely on manual monitoring,this is not only a waste of human resources,but also appears accidents such as missed,wrong seizure events and other.Therefore,the study of intelligent traffic monitoring system is of great importance.This paper studies the target detection algorithm based on Gaussian mixture model,introduced pretreatment target detection process sequence of images,segmentation,feature extraction and object recognition in four steps.On the basis of this work,we improve and optimize the gaussian mixture model,and the improved algorithm is applied to the intelligent transportation monitoring system.The main work is as follows:First this paper introduces the moving target detection algorithm based on video research background and research significance,analyzes the main problems of current target detection technology.And detailed introduction of several typical motion target detection algorithm and compared to the performance of these algorithms.Analyze the gaussian mixture model of classical background modeling algorithm,analyzed how to use gaussian mixture model to model the background,background model initialization,extracting motion prospect in video image information,analyzed how to use mathematical morphology method to processing prospect of information.Because of modeling large amount of gaussian mixture model,the algorithm complexity is high,and can't meet the problem of real-time video monitoring,a improved model based on super pixel clustering blocking gaussian mixture model is proposed.The model processing pixels by areas,SLIC algorithm is utilized to extract ultra pixels instead of single pixel points of a video frame as the foundation of the gaussian mixture model.Using Matlab simulation experiment results show that the improved Gaussian mixture Gaussian mixture model proposed ratio faster processing speed.Finally,we apply improved algorithm to intelligent traffic monitoring system,designed the basic architecture of intelligent traffic monitoring system,Set up the traffic video monitoring system of the hardware equipment,concrete realization of the project in each module.Using the system to test the vehicles in the road,the test results showed that the system can quickly to correct into the scope of video vehicle detection.