标题：For the revision of GIS data by means of digital images correspondence of image features and GIS objects is essential. In case of small scales this is complicated due to the fact that many GIS data have been recorded by digitizing maps. Thus, geometry of generalised maps has been transferred to the GIS. Hence, geometry of geocoded orthorectified satellite images is more accurate than small-scale GIS data. The main goal of this paper is to offer a contribution to the .ow of information between GIS and image analysis which makes it feasible to utilize the GIS data as prior information for image analysis on the one hand and realize quality control or revision of GIS data using digital images on the other hand. Both the GIS objects and the image features are given as vector data, but raster representations can be derived easily, so that we may benefit from the advantages of the specific representation. Since we assume that the satellite image and the GIS data have been approximately matched interactively, searching space for the corresponding image features is defined by a maximal distance to the GIS object. Therefore, we apply a local search strategy to establish correspondences. As image features we are using lines that have been extracted from satellite images and which have been classified as roads due to their spectral characteristics and length. The performance of the method is demonstrated using IRS-1C data as examples
期刊名称：ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
印刷版ISSN：2194-9042
电子版ISSN：2194-9050
出版年度：2000
卷号：XXXIII Part B2
页码：97-104
出版社：Copernicus Publications
摘要：For the revision of GIS data by means of digital images correspondence of image features and GIS objects is essential. In case of small scales this is complicated due to the fact that many GIS data have been recorded by digitizing maps. Thus, geometry of generalised maps has been transferred to the GIS. Hence, geometry of geocoded orthorectified satellite images is more accurate than small-scale GIS data. The main goal of this paper is to offer a contribution to the .ow of information between GIS and image analysis which makes it feasible to utilize the GIS data as prior information for image analysis on the one hand and realize quality control or revision of GIS data using digital images on the other hand. Both the GIS objects and the image features are given as vector data, but raster representations can be derived easily, so that we may benefit from the advantages of the specific representation. Since we assume that the satellite image and the GIS data have been approximately matched interactively, searching space for the corresponding image features is defined by a maximal distance to the GIS object. Therefore, we apply a local search strategy to establish correspondences. As image features we are using lines that have been extracted from satellite images and which have been classified as roads due to their spectral characteristics and length. The performance of the method is demonstrated using IRS-1C data as examples
