Research on Detecting and Tracking the Moving Objects
|School||Henan University of Science and Technology|
|Course||Applied Computer Technology|
|Keywords||Moving object detection Time information and Space information Multimode models Kalman Filter|
Object detecting and tracking is a challenging subject within the field of computer vision. The common use of object detecting and tracking is in the task of video coding, intelligent traffic, surveillance,image retrieval, military industry and so on.In this paper we will research the way of detecting moving object based on time information in combination with spatial information and the way on how to track the moving object based on theory of Mean Shift.As to moving object detection,the paper firstly discusses three traditional algorithms including Consecutive frames difference, Background subtraction and Optical flow,then the paper points out advantages and disadvantages of each algorithm.The algorithm of Consecutive frames difference is very simple that we just need to do substraction beetween to frames, the moving object we detect by this algorithm often contains some holes in it ; The second algorithm needs to construct the model of background,but the model is not easy to construct .The last algorithm needs to calculate the optical flow field that it can not meet the need of real-time.after that the paper puts forward a new algorithm which is based on time information and spatial information.The new algorithm well makes up the defects of traditional algorithm,consequently it improves quality and stability of traditional algorithms.0n moving object tracking,the paper focuses on enhancing stability of the traditional agorithm which is based on theory of Mean Shift.To achieve this goal,we firstly construct several different models according to properties of the moving object,then we can get the most proper model to track moving object depending on comparing similarity between moving object and background under different models.In addition,to track the object with high speed,first of all,we use Kalman Filter to predict the possible position which the object may move to with high possibility,then we can use Mean Shift to calculate the object’s exact position from the object’s possible position.