Research Of Similar Trajectory Query Based On The Moving Object Data Stream

In recent years, with the further promotion of the Internet of things technology and sensor technology, all kinds of intelligent transportation applications based on a wide range of monitoring vehicle data got rapid development. Similar trajectory query is one of important applications. It can develop the similarity relation of the trajectories of vehicles based on the existing vehicle monitoring data. This characteristic provides support for the business that related to transportation. Such as police to deal with cases that are related to vehicles, track and analyze criminal suspect vehicles and urban road planning. So it has important significance and practical application value.Under the background of monitoring urban vehicles, this paper carried out research work according to the problem of similar trajectory query under the massive license plate data set in view of the moving urban vehicles. On the basic of existing research achievements, we focused on the judgment of similar trajectory of the data that come from the recognition of license plate and the process of similar trajectory query that is distributed with huge amounts of data set. The main work of the paper is as follows.First of all, considering the license plate recognition data has the features of associated monitoring points and associated in sequence with a flow way, we define the method of querying vehicle similar trajectory based on the under the license plate recognition data set.Second, we proposed a similar trajectory query computing technology based on multistage Map Reduce pipeline, and then developed the corresponding similar trajectory query system based on Hadoop. We optimized the method from the aspect of filtering redundant data, division data and performed in parallel and the definition of track similarity. The technology achieved similar trajectory query based on mass license plate recognition data efficiently in distributed environment.Last, we designed a set of experiments which is based on massive real license plate recognition data set. According to the experiment, we verified that our work has the feature of high performance, scalability and accuracy.