The possibilities to Support ZB GIS database update using object-based image analysis in ecognition developer software

Abstract

The process of updating a spatial database is a necessary part of database management in order to keep the stored information of an acceptable quality. The first step of database update requires change detection. Many methods have been suggested to detect changes, mostly pixel-based. Recently, with the spread of very high resolution images and object-based image analysis, object-based methods were developed, too. This article presents object-based change detection method for update of the vector database ZB GIS, that is a geometric base of the Slovak National infrastructure of spatial information, using orthophoto of the area of interest. The method stages include the following: segmentation of orthophoto using the geometry of database objects and further according to spectral and spatial information, classification according to ZB GISR, defining reclassification rules between two classes. The proposed method was tested in two localities – Malženice (agricultural landscape) and Chopok-Jasna (mountainous landscape) reaching the overall accuracy of classification 87.12% and 84.55%, respectively. The main limitation of the method is that it can be applied only for polygonal objects.