Research On Application Of Object Detecting And Tracking In Changing Background

Varied background refers to the background of the object in video imagessequence is dynamic and variable. Object detection and tracking system, thebackground of the traditional relatively static or background change but is relativelysimple, only treated as noise, and in the reality background changes on the objectdetection and tracking is not to be ignored, the influence of the traditional methodsand even may cause the loss of the object. Varied background research on objectdetection and tracking is the basis in the field of computer vision and one of verychallenging issues has important practical significance and theoretical value, at thesame time in the industry, national defense and other fields have broad applicationprospects.For object detection and object tracking in this paper is divided into two parts, thefirst of several typical object detection are analyzed and compared, and mainly studiesthe AdaBoost algorithm, and to the need of the reality, to traverse the AdaBoostalgorithm parameter scope has carried on the limits, make the window scalingcoefficients at the same time can be adjusted adaptive. The experimental results showthat, this paper USES the algorithm is concise, efficient, able to quickly andefficiently detect the moving objects.Object tracking part of several kinds of most commonly used in the mainstreamof object tracking algorithm was analyzed, finally selects the particle filter trackingalgorithm to detect the object. On particle degradation problems, based on multiplefeatures fusion particle filter, introducing the Mean Shift algorithm, with less number of particles achieve better tracking results.Designation of this paper has a wide scope of application of object detection andtracking system, can be used for all kinds of rigid and non-rigid object, such as humanface, body, vehicles and objects, etc. Due to the limitation of space, this paper only inthe most popular human face and artificial object as experimental subjects werestudied, the other. The same applies to all kinds of object. Finally this paper willintegrate the object detection and object tracking technology, to complete a set ofchanges under the background of the object detection and tracking system design, hascertain theory significance and the practical reference value.