A Integration Method For Change Detection And Change Type Determination Based On Decision Tree Rules

Land-use map is the foundation for region programming and environment research, and the precondition of all kinds of land management. Now, the high-speed development of economic accelerates the speed of land-use change. So, there have very important signification to update the land-use map exactly and provide the exact and uptodate land-use data for governments.After the study of the use of "3 s" in the land-use map updating, we evaluate the method of land-use map updating from three aspects, the data sources, change detection and map updating technology, and presents a integration method for change detection and map updating based on decision tree rules. Knowledge discovery and decision rules are applied to realize the feature layered extraction, based on the knowledge base, comprised by both spectral information and geo-information, such as altitude, texture, geometry, mining from the new remote sensing data, old land use map and auxiliary information. DEM and slope map. Then, operate the feature extraction imagery and the feature vector map to detect the change information and determine the type. Experimental results and some practical applications show the efficiency of this method. Compared with the change detection based on the value of the spots, this method have the superiority in precision of change detection and map updating, and the knowledge base, which contains more information of image, has been used well. So, this method can be used in map updating.