Automated building simplification using a recursive approach

This article presents a new generalization algorithm for automated or semi automated buildings simplification based on the recursive approach. From the data set with a higher level of detail another data set that containing a lower level of detail can be derived. The proposed algorithm has an ability to detect and simplify buildings rotated in random position and does not depend on ordering of points. To determine the angle of rotation ' of the building, a smallest area enclosing rectangle is constructed. The splitting procedure is performed on the basis of splitting criterion calculated for each edge of the building. The simplified edges are replaced with regression lines, whose parameters are determined using the least squares method. The algorithm was implemented in C++. With the use of ArcObjects libraries, a new DLL application designed for ArcGIS 9.x was created. This tool is applicable to the polygonal shape files.