Urban Change Detection on Satellite Images Series. Application to Catalunya Area

In this paper, a methodology developed for the detection of significant changes for cartographic purposes on a set of satellite images is applied to a series of three SPOT images. The detection of territorial changes is a keystone for the assistance to the scheduling of the cartographic updating projects. For this reason the Institut Cartografic de Catalunya (ICC) develops a project aimed to the detection of actual territorial changes on satellite images. A second objective of the project is the analysis of the territory dynamics. The main variables to be obtained in the process are not static land use maps corresponding to each of the pictures in the series but the speed of territorial change in the area. This output is especially interesting on the areas where these changes are more frequent, basically around the cities and along the sea side. Main input data used in the project are SPOT images obtained twice a year for change detection. The initial area of interest is the metropolitan area around Barcelona. As a previous step, a series of atmospheric correction processes are applied to each image and covers that are prone to generate blunder changes are detected and marked off. Next, multitemporal RGB images with information from two different SPOT pan bands are generated. Those images are processed by a combination of adapted threshold levels and photo interpretation analysis. The procedure is a tree classification system that allows detection and confirmation of the changes with the following images. General minimum mapping unit is 1 ha, but it is reduced for urban and other interesting cases. The final results showed good accuracy on detection of the actual changes only with SPOT imagery. They also indicated that the amount of changes for a year ranges between 0.56 and 1.13% of the studied area, with a mean value of 0.88% of the territory. Moreover, the detection methodology was found robust and it was difficult that a real change went unnoticed.