A high-precision optimization method for multiclass classification building contours
Aiming at the problem that existing conventional optimization methods for building contours usually do not consider the topological relationship between adjacent contours,which leads to the existence of cross-topological errors in adjacent building contours segments after optimization,thus affecting further work,this paper proposes a high-precision optimization method for multiclass classification of building contours.The method of R-tree spatial indexing and classification optimization of building contours is adopted to realize the rapid optimization of building contours of different categories and maintain the topological relationship between the contours,and the building contours after preliminary optimization are further adjusted by adding intersection and intersection over union constraints and Hausdorff distance,which can effectively solve the problems of over-optimization of the contours and too many details,and obtain the building contours close to the true value.The experimental results based on contour optimization show that the method proposed in this paper is able to solve the cross-topological errors problem in adjacent building contours segments after optimization,and the results of the study have certain reference significance to the field of building contour optimization.
multiclass classificationcontour optimizationintersection over union constraintR-tree spatial indexHausdorff distance
万方数据
维普
