Similarity Analysis And Anomaly Detection Technology Based On Trajectory Motion

At present,in the field of video surveillance,the processing of trajectory data has been realized by clustering and anomaly detection.Various clustering algorithms have their own advantages,this paper discussed the clustering and outlier detection algorithm,and has analyzed the advantages and disadvantages of common types,such as partition method,hierarchical method,density based method,grid based method and modelbased method.Then the potential solutions and some improvement of the algorithms are pointed out.some algorithm may be combined,so that the final result can be improved.This paper presents a new distance measurement,discusses the different distance measurement of the particular calculation process,the characteristics of each of these methods.This paper has discussed the distance measurement,the detailed application of them and elaborated the distance measurement method of the trajectory similarity,through the way of possible improvements,which is more suitable for the distance.So the similarity between different trajectories can be measured more accurate.In order to improve the method of calculating trajectory similarity,the Euclidean distance has been improved,which makes the measurement results of trajectory clustering and anomaly detection more correct.At the same time,it avoids the lack of the similarity measurement of complete trajectory information.At the same time,from the point of view of anomaly detection,there are some difficulties in the trajectory data.This paper discusses the information theory tools in detail,as well as the employment of the method.The paper proposes the method of information theory tools used in trajectory anomaly detection,through further digging out the trajectory data in motion angle information.The information tools can analyze the trajectory similarity with high level similarity,to detect the hidden abnormal trajectory in the normal trajectory data.In this paper,the common clustering algorithm and anomaly detection algorithm are analyzed,and the improved scheme proposed in this paper can further improve the effect of the original algorithm.Finally,the experimental results are compared with the results of different algorithms and schemes.