The Change Detection Of Road Network Map Data Based On Remote Sensing Image

In recent years, photogrammetry and remote sensing has developed at very high speed, remote sensing data increases by degrees , now it has become an important data source of GIS and means of data updating. Comparatively, GIS data has been assistant information for the processing of RS data, using in automatic extract of the information of semantic meaning and non-semantic meaning. The integration of GIS and RS is mainly used for change detection and real time updating, along with the fast development of RS technique ,the integration is more and more limited by antinomy of fast acquired, substantive images data and laggard means of data processing. Substantive images and correlative data and means of processing, mainly extract information we needed, has tremendous contrast. Fast innovation of GIS data has become a challenge to the field of photogrammetry, remote sensing and GIS.Change detection is an important premise of updating of GIS data. This paper attempts to detect the change of road data in old vector using new RS data automatically or semi-automatically. Based on the automatically and accurately registration of RS image and road network vector map, we investigate high roboticized and accurate detection methods aimed at two kind of change of road: One is, base on old map road network, detect there whether or not has relatively changes in roads on image. In this process, author defines buffer distance, adopt method of multi-scale template matching and strategy of knowledge judgement to detect whether there has changed; The second is, base on new RS image, detect newly added roads automatically and semi-automatically. here mainly study projection track model and LSB-Snake model, and improve them respectively for the need of paper. From the point of view of practicability and by comparative, this paper adopt improved LSB-Snake model to detect new added road network.By reforming such strategy, we do experiments using several different scale images, the experiments indicate that, detection of the change of road network is high accurate, and improves detection robotization.