Research On Change Detection And Update Of Surface Elements In DLG Database

The fundamental geographic information database is an important part of the spatial data infrastructure. It is a significant fundamental subsystem in the economic information system, and provides basic geographic data for various geographic information application industries. As the basis of spatial analysis, DLG is an important spatial data type in the fundamental geographic information database,which is a vector expression of natural and humanistic phenomena. In order to keep the timeliness of DLG, it must be updated constantly. Surface elements are important features in the DLG database, especially in large scale datasets reflecting urban areas,such as buildings, green land, lakes, etc. Some of these objects like buildings change fast, so this paper focuses on the matching and update technical methods of surface elements. Compared with the version update, DLG database incremental update is cost-effective but more complex. It needs to match homologous elements, detects and classifies the factor changes, and incremental updates the database. Therefore, this paper researches the technical method for DLG database change detection and update.The main research contents are as follows:(1) The automatic matching of surface elements has been analyzed comprehensively,and the existing methods have been classified and summarized completely.Aiming at the problem that most of the current methods can't deal with the non-one-to-one surface elements matching, this paper extracts multiple feature vectors of the corresponding surface element as the matching basis, and regards the different characteristics as input, then uses the logistic regression model to realize the automatic matching of the surface elements. Building matching is taken as the example, and the experimental results show that the proposed method can effectively solve the one - to - many, many - to - one and many - to - many matching relationships between surface elements.(2) Aiming at the problem of detecting and classifying feature changes, this paper constructs a rule-chain based classification model. It regards the decision tree model as a guidance and follows the variation rules between elements. When determining the change type of specific ground elements, main differences between elements are adjusted firstly. Then, the difference feature vector is recalculated and substituted into the matching model. According to the matching result and features, changes of surface elements are detected and classified by certain rules and the change type of surface elements are marked finally.Experiments show that the proposed method can detect the change of surface elements more accurate, and distinguish change types of surface elements.(3) On the basis of the detection and classification of surface element changes, this paper develops different updating operations according to the change types, and realizes the automatic updating of surface elements. In order to manage the historical data and the relationship of the data before and after changes, new storage models are designed for different change types. By establishing a permanent unique ID and adding a time stamp, the dataset establishes associations for the changing elements, which is convenient for further spatiotemporal analysis.(4) Based on the theories and methods of this paper,the prototype system of surface elements change detection and update in DLG database is designed and developed.The system can match homologous surface elements and detect and classify changes of the element. This system can also extract the change information,update the base of current situation incrementally and save the historical data after the change classification.