THE QUALITY CONTROL OF VECTOR MAP DATA Wu Fanghua Liu Pingzhi Jincheng Xi an Research Institute of Surveying and Mapping (P.R.China ShanXi Xi an Middle 1 Yanta Road 710054) (e-mail :wufh999@yahoo.com.cn) Abstract The basic circumstances of producing vector map data are introduced in this paper. And the methods and measures of data quality control in the course of data production are discussed. The data quality control is considered as a synthetical system engineering. And it needs not only strict regulations, but also suitable quality standards and software systems for checking and controlling the quality. According to data producing methods, data quality control system is made up of features-updating quality control subsystem, features-capturing quality control subsystem, data quality checking and accepting subsystem, and data quality evaluation subsystem. Key words fundamental geographic database; quality control; data quality evaluation 1. INTRODUCTION Data quality is the life of map databases. Quality control of map data is a very important issue for the digital information engineering of surveying and mapping. It integrates management with technology. Quality control is a difficult and complicated systematic engineering especially for large spatial databases. For different ways of production, the contents and methods of quality control are quite different. It is of great significance to study how to carry out quality control in the process of data acquisition and [1,6] database establishment and to ensure the quality of fundamental geographic data. Presently the establishment of 1:50000 fundamental geographic database is being carried out throughout the country. This geographic database is the foundation for establishing the spatial data frame of Digital China and the carrier for various thematic information. The 1:50000 fundamental geographic database is different from other kinds of GISs in their features and requirements. It is a non-centralized database system and consists of multiple databases which are correlated but independent of each other in operation, including the vector map database, the pixel map database, the elevation database, the digital orthographic image database and the geographic name database. These five databases have the similar fundamental information contents, functions and products modes, as well as unified classification codes and quality standards. Among them the vector map database is the one consists of topographic features such as drainage, contour, boundary, traffic and habitation represented with vector data. It includes the spatial relationship and the attributive information of topographic features. The vector map database is organized according to specific rules and based on different sheets and layers, and its data is encoded according to 1
the encoding standards for features classification. The quality requirements, controlling measures and the focuses of the five databases are different from each other. Taking the production of 1:50000 digital vector maps as an example, in this paper we will introduce the quality control technology, methods and steps for digital maps production. 2. PRODUCTION OF THE DIGITAL VECTOR MAPS The digital vector maps are the vector data sets of fundamental features of existing topographic maps, keeping the spatial relationship between the features as well as the corresponding attributive information. They represent the ground objects quite comprehensively and consist of several features. Each feature represents its geometric properties by using points, lines and areas. At the same time, the feature is assigned attributes and divided into several data layers according to the geographic features classification. The production flowchart is shown in figure 1. Topographic map or aerial photograph Satellite image Data acquisition Data extraction Data acquision with GPS Initial vector data Vector data of land coverage Vector data of road net Data fusion, graph editing, processing of features relation, data layering, data overlapping Quality control Digital vector map Figure 1.Flowchart of digital vector map There are multiple methods of data acquisition for 1:50000 digital vector maps. Data acquisition includes direct acquisition methods, such as digitized field survey, mapping with aerial photographs, mapping with image data and road data survey with GPS, and indirect acquisition methods. In our operation we mainly acquire data by scanning and vectorizing the newly produced or revised topographic maps. The diversity of the methods for data acquisition results in the diversity and complexity of the contents of data quality control. Quality control must be carried out throughout the process of production. 3. THE CONTENTS OF STUDY ON DATA QUALITY CONTROL OF THE VECTOR MAPS According to the data production procedures of the vector maps, the quality control 2
system of map data consists of the quality control subsystem of features updating, the quality control subsystem of features acquisition, the data quality checking subsystem and the data evaluation subsystem. The function of the quality control subsystem of features updating is embedded into the features updating system. The function of the quality control subsystem of features acquisition is embedded into the features acquisition system. Specifically the contents of the study include the following five aspects [ 3]. (1) Quality control of features updating The quality control of features updating includes the situation check of map data, quick updating of topographic map data, automatic generation of topologic relationship of the updated features, quality check of the updated data, and the overall quality evaluation of the updated data. (2) Quality control of features acquisition The quality control of features acquisition includes quality control in the operation of features acquisition, quality control flow of features acquisition, and data quality check of features acquisition. Among them the first is the focus of the study. And the third is mainly to check up the results of the subsystem according to data quality and make adjustments according to the operation flow. (3) Quality check of map data The quality check of map data includes nine aspects such as integrality check of the data, logical consistency check of the data, position precision check of the data, attributive correctness check of the data, time correctness check of the data, data overlapping check, topologic relationship check of the features, spatial relationship check of the features and the meta-data check. (4) Study on data quality standards Quality standards are the foundation of quality control. The corresponding standards and specifications are the bases of quality analysis and evaluation. To carry out computer processing needs to model the standards and specifications and transfer the models to a series of rules recognizable to computers. (5) Study on data quality evaluation and the evaluation models [5] The evaluation models determine how to evaluate data quality in a region or in a sheet of map. They are the bases of ensuring that the evaluation results are objective and correct. The precondition of establishing a satisfactory evaluation model is to analyze and quantify various specifications and standards for vector maps production, to master the experiences of carrying out these specifications and standards, to cooperate with the 3
experts engaging in practical operation, and to gradually modify and optimize the control model after evaluating data quality of the proof map. Quality evaluation needs to provide not only quantitative parameters such as wrongs, omissions and errors, but also qualitative conclusions of whether the map sheet is qualified or not. 4. RULES AND REGULATIONS FOR QUALITY CONTROL OF 1:50000 VECTOR MAP DATABASE Quality control must be carried out throughout the process of database establishment. For every stage and every step, there must be the corresponding technical specifications for quality control and the corresponding quality examiners and supervisors to carry out the measures of quality management. To guarantee the quality of the data products, quality examination must be carried out strictly in the process of 1:50000 vector maps production and before the final products are subjected to the central database. The quality of the production procedures must undergo three-level examinations (including (a) check and verification; (b) check and acceptance; (c) higher-level check and acceptance) carried out by the production institution. Data production includes the procedures of local revision by means of aerial photogrammetry and topographic maps vectorization by means of scanning. Both the two procedures must undergo the examinations (a) and (b). After the two examinations, the products of local revision by means of aerial photogrammetry are turned over to the procedure of topographic maps vectorization, and the final products of topographic maps vectorization are turned over to examination (c). All the three-level examinations are carried out with the sheet-by-sheet method. One hundred percent of the map sheets must undergo the examinations (a) and (b). Examination (c) can be carried out to some or all the products depending on the specific conditions such as the difficulty of the task, the amount of the documents and the method of operation. At least 30% of the whole sheets must undergo examination (c). All the three examinations must be carried out independently and cannot be omitted or replaced. All the results of the three examinations must be recorded. The qualified products having passed examination (c) must be subjected to the central database and undergo the examination for entrance. The examination for entrance is carried out in batches, including documents check, products sampling and entrance check. Products sampling is to select samples from a batch of map sheets according to the specific ratio (based on the number of map sheets of the batch). And the selected samples must undergo the examination for entrance. The examination for entrance will not finish until all the data have correctly entered the central database. 4
5.TECHNOLOGY OF QUALITY CONTROL The technology used for map quality control mainly includes quality control methods, error analysis and processing models, automatic examination methods and quality evaluation models. 5.1 Quality control methods for map data Quality control of map data means controlling the quality of map data in the process of features updating and acquisition. The following methods can heighten the automatic level of quality control. (1) Templates matching The control templates are designed on the basis of the error control models. In the process of features updating and acquisition, the attributive data are matched with the control templates, the data failed to meet the requirements of the control templates are controlled automatically. Thus the attributive correctness is guaranteed. (2) Precision control Precision control in the process of features updating and acquisition plays a decisive role in the overall precision control of map data. After the acquisition of map features, the errors of map data can only be adjusted and the improvement of data precision is limited. Through the errors analysis and processing models for map data, both the original maps and the updated data undergo the procedures of error processing and precision [4,7] control. Thus the updated data and the acquired data can meet the requirements of precision. (3) Fast error location The function of the examination module of the data examination subsystem is embedded into the features updating and acquisition system. The data are classified and then undergo the corresponding real-time examination. The examination results are correlated with the updated data and the acquired data. Quickly and automatically the errors are located and visualized, thus they can be processed conveniently. (4) Overlaying of multi-source data Multiple kinds of documents and data such as orthographic images, original maps, the data updated with GPS, and the data of the topographic names are overlaid transparently or semi-transparently in order to check the correctness of time and correlation, as well as the precision of spatial position. 5
Figure 2. Overlaying of muiti-source data (5) Automatic features overlapping Based on the attributive consistency of features in adjacent sheets and the specified precision of spatial position, the features in adjacent sheets are overlapped automatically. 5.2 Error analysis and processing models for map data The error analysis and processing models for map data include the spatial position error analysis models, the attributive mistakenness analysis models, and the error processing and control models. (1) Error analysis of spatial position [2] The spatial position errors of map data are generally classified to two types, i.e. the intrinsic errors of the original maps and the introduced errors in the process of digitization. For the convenience of processing, the errors of the two types are divided into the random errors, the systematic errors and the abnormal errors based on their influences. And the corresponding error processing models are established respectively. (2) Attributive mistakenness analysis On the basis of attributive definition domain such as code, parameter and representation, the attributive mistakenness is classified and its properties and influences are ascertained. Then the mistakenness is turned over to the examination and processing system of map data. (3) Error processing and control models The error processing and control models consist of the error processing models and the error control models. The error processing models are mainly to process the spatial position errors. The random errors, the systematic errors and the abnormal errors are different in the properties of their influences, so the models and methods for error processing are different. For the random errors, because of the randomness of the single influence and the compensation property of the overall influences, they can be controlled with high precision coordinates of the control points, and their 6
influences can be weakened by rectifying the errors with least square adjustment. The systematic errors can be rectified and controlled with specific function models (such as polynomial approximation). The abnormal errors are generally removed by abnormality examination or robust estimation. According to the relevant specifications and the properties of the feature, a control domain (consisting of definition domain, logical consistency domain and integrality domain) for the corresponding feature is set up. Based on the control domains for all the features, the error control models can be established. 5.3 Data examination methods (1) Automatic examination On the basis of the data error processing and control models, the method for the automatic examination of data integrality, logical consistency, position precision, time correctness, topologic relationship and meta-data is designed. Then the map data can be examined automatically. Figure 3. Automatic quality examination (2) Semiautomatic examination The method for the semi-automatic examination of attributive correctness, data overlapping and spatial relationship of features is designed. Then the map data can be examined semi-automatically. Figure 4. Semiautomatic quality examination (3) Fast error location The examination results of map data are automatically correlated with map data. The errors are quickly located and visualized, so that the data can be quickly edited and modified. 7
Figure 5. Fast error location (4) Examination of the updated data The orthographic images and the original maps are overlaid over the map data transparently or semi-transparently. Thus the correctness of time, correlation, and the precision of spatial position of the data can be examined. Figure 6. Non-updated roads (5) Examination of the symbolized maps The digital maps are symbolized and printed out. Then they are compared with the original maps to examine the omissions, the correlativity, the precision of spatial position of the data, as well as the layout of the printed sheets. (6) Quality evaluation The primary conclusions on the quality of the map sheets are drawn based on the results of the examination. Figure 7. Primary conclusions on map quality 6.CONCLUDING REMARKS 8
The quality control of map data is mainly to study the quality control technology for the digital vector maps of series of scales. Taking the production of 1:50000 digital vector maps as an important example and based on the properties of digital map data, we lay down a set of quality control standards and quality control procedures suitable for production. We also work out a set of software for quality control, check and acceptance on the basis of the codes, the data models and the data formats of national fundamental geographic information. With regard to the aspects such as the position, the attribute, the topology and the logic consistency of the map data, we inquire into the rules and regulations, the bases and the contents of quality control in the process of vector maps production, and put forward the methods for improving the quality of the digital vector maps. With the heightening of the requirements for map data quality, the importance of the quality of map data is recognized more deeply than before. The quality control of map data will be one of the study subjects in GIS field in quite a long time. Data producers need to work hard to improve the quality of map data. References 1. Wu Fanghua, Quality control of the establishment of 1:50000 vector map database, Journal of Xi an Research Institute of Surveying and Mapping, 2002.4 2. Tong Xiaohua, Shi Wenzhong, Liu Dajie, Errors distribution, check and processing of the digitized data in GIS, Journal of Wu han University of Science and Technology, 2002.2 3. Wu Fanghua, Theory and practice of quality control for vector map data. Zheng zhou University of Information Engineering, Dissertation for Master s degree, 2002.4 4. Tong Xiaohua,Liu Dajie,A Methodology to Adijust Digital Cadastral Areas in GIS, The 20 th International Cartographic Conference. ICC 2001 Beijing China.volume 5; 2866~2874 5. Nina M.Kelly, 2000, Spatial Accuracy Assessment of Wetland Permit Data, Cartography and Geographic Information Science, Vol.27, p117~127 6. Goodchild,M.F. and Gopal,S.(ed.),1989b,the Accuracy of Spatial Databases. New York, Taylor and Francis 7. Goodchild,M.F. and Hunter,G.J.,A Simple Positional Accuracy Measure for Linear Features. International Journal of Geographical Information Science. Val.11 No.3, 1997:397~408 8. Goodchild,M.F.,1998, Geomatics and geographic information science: tran-century directions and application,proceeding of SIST 98, Wuhan, China: 70;304~311 9