Detection And Tracking Of Moving Target Based On Image Sequences

Detection and tracking of moving target is one of mostly important questions in the area of computer vision, and broadly used in areas such as video surveillance, medical diagnoses, intelligent traffic, movement analyzing,so it is very significant and with great practical value.Method of moving target detection and tracking is mostly researched and tested in this paper. Firstly, the method of moving target detection in common use at static scene are discussed, on the basis of foregoing discussion, improved and more scientific applied means is adopted. Then, method of moving target detection based on an arithmetic which combines object's centroid with prediction method is researched. Stability and robustness of target tracking is improved.Objective of moving target detection is to pick-up changing extent from image sequences in the background. However, because of background image's dynamic transformation, such as weather, sunshine and shake, moving target detecting becomes a very difficult task. Moving target detection method based on inter-frame difference algorithm and subtraction of background is mainly researched in this paper. Aiming at advantage and disadvantage of inter-frame difference algorithm and subtraction of background, this paper combines two methods, brings a improved detection method, too hypersensitive with dynamic scene'changing, which is disadvantage of subtraction of background method, is eliminated. At the same time, more comprehensive communication about moving object can be obtained. Then, subsequent disposal like mathematical morphology is taken to get more accurate target.Moving target tracking is to determine an object's locality in different frames in image sequences. Object'planar coordinate in every frame is confirmed by analyzing image sequences. Then the object in different frames in images sequence is parallel connected with different eigenvalue. So the whole contrail of this object is gained. In fact, target tracking is to estimate object's moving track. Method of target tracking based on an arithmetic which combines object's centroid with forecasting method is adopted in this paper. Even the object is hided for a while; its next imaginable location in a while can also be predicted with its previous location information by this method. So, when the object appears again, it still can be steadily tracked. Feasibility of this arithmetic is validated by a series of experiments which take stimulant aeroplane's flight scene as object.