|
|
|
The Design of Massive Image Database Management System
Pai-Hui Hsu
Associate Research Fellow
National Science and Technology Center for Disaster Reduction,
Taiwan
Email: hsuph@ntu.edu.tw
Liang-Hwei Lee
Department of Civil Engineering
National Kaohsiung University of Applied Sciences
Email: lhlee@cc.kuas.edu.tw
Fong-Gee Yiu
Department of Civil Engineering
Lan Yang Institute of Technology
Email: gee@mail.fit.edu.tw
Recently, the remote sensing system has been developed towards the tendency of multi-sensor, high-spatial, high-spectral and high-temporal resolutions. The consequence is that the abundant remotely sensed data provides us with rich information for the investigation of earth surface, environment and global change helpfully. However, the huge image data has also built up the limitations when they are applied to different research and practical purposes. Especially for the image data sharing and delivering via the internet technology, the massive data would be an obstacle for image management including the image storing, querying, transmission, and displaying. Therefore, how to manage the huge image data effectively using the advanced informant technologies will be a critical issue in practical applications.In this paper, a web-based image database management system was developed for querying, analyzing and displaying huge image data sets such as the high resolution satellite images, digital aerial photos, and high-resolution digital terrain models (DTM). In order to access the huge image data efficiently, two new file formats for image compression, named Huge Image File (HIF) and Internet Image File (IIF), are designed based on the concept of image pyramid with multi-resolution data structure. The HIF file consists of the compressed image data using wavelet and additional spatial indices which can be used to decompressed the partial image rapidly. Unlike the HIF, the file structure of IIF file itself is a kind of database whose data structure are based on the quartree. The massive image can be decompressed rapidly using the spatial indices stored in the nodes of the quadtree. One of the advantages is that there is no limitation in the file size of the IIF image.Several database technologies including data processing, retrieval and sorting techniques are also included in the design of the massive image database management system. In addition, this system itself is a platform for developing Web-GIS applications. The user can effortlessly browse and query the 2D images or 3D terrain information via the internet browser arbitrarily. This system also supports the loading balance for multi-user requests. Furthermore, the system also provides software development kit (SDK) and customized components for system developers to create new functionalities for different applications.In order to test the performance of the massive image database management system, a huge seamless image of 2 Terabytes (consisted of 4000 aerial images) and a high-resolution DTM data set with 5-meter intervals were used to display the 2D images and the 3D terrain model real time. The experiments showed that the performance of the massive image database system is satisfactory in most circumstances.
Keywords: Massive Images, Web-GIS, Database Management System
|
|
|