Abstract: Remote sensing images are the important information,source in forestry system. Forestry researches are vast in territory,scene are large and data are huge. And generally it requires images at multiple resolutions from 1 to hundreds meters. So,it's crucial to make a solution which can show high performance in querying and processing,provide strong support on storing and managing the huge amount of multi-resolution and multi-band image data,and browsing in 3D view. To meet those demands,based on Oracle 10g GeoRaster component,the method of designing image library was discussed. The feasible scheme of images storage and organization of image library were put forward. In this study,the organizing method of image layers was divided into seamless image layer model and tiled image layer model. Different approach has been carried out for both models in building spatial indexing,I.e. building blocks index for seamless image layer model and tiles index for tiled image layer model,respectively. The PL/SQL codes of building pyramid for GeoRaster objects were listed. Zip and JPEG 2000 compression formats were applied to improve the performance of data extracting. Finally,a prototype system for the design solution was introduced.

Key words: image library, Oracle GeoRaster, image pyramid, image compression