Target Trajectory Analysis Based On Self-organizing Neural Network

With the development of society,automatic video trajectory with the video image data base analysis has important significance in various fields,including the extraction of video image information,data information,establishing the data model suitable for inspection.A large number of video data is the data base of the movement track of the object,using the method of digital image processing,filtering and extracting the background image to extract the motion trajectory,trajectory of object related research needs,is the basis for the analysis of moving object trajectory.The main feature of trajectory data is extracted by effective method.The trajectory is clustered and the motion characteristics of trajectory are analyzed.Firstly,this paper introduces the research background and research status of moving target trajectory,and then introduces the common methods of extracting moving target trajectory from video,and analyzes the advantages and disadvantages of each extraction method.Next to the short track trajectory data,using the improved grey Markov forecasting method,the completion of the track,forecast the movement trend of the future trajectory of short time.Next,the manifold methods of manifold learning are introduced,and the characteristics of applicable data sets of different methods are analyzed.According to the classical Euclidean distance metric lack the trajectory feature description,can not accurately describe the difference between the two-dimensional trajectory manifold characteristics,this paper will track map data into vector field,establish the trajectory data of vector field,retains the main features of information visualization,trajectory data can be said that the shape feature trajectory data directly express.Finally,the self-organizing neural networks and kmeans clustering are introduced.The effectiveness of the proposed algorithm is verified by experiments using hurricane data in the Atlantic.