Abstract:

The feature extraction quality and accuracy of the fine modeling building plays an important role in the feature information extraction of buildings clearly and accurately.The traditional vector estimation method is affected by noise,and exists some misjudgment problem,this paper proposed a feature extraction method of the point cloud mobile minimum two multiplication of buildings based on vector estimation.This method firstly uses the moving least squares method for estimation of normal vector,then the mean K neighborhood method as a significant index vector included angle point feature identification,finally the extracted feature points set by resampling,to eliminate redundant information.The experimental results show that the estimation of point cloud normal is more accurate and robust by moving least square method,thus effectively improve the accuracy and reliability of the building point cloud feature extraction,the feature point set resampling can delete the redundant feature point,the feature line extraction is more concise,clear and complete.

Key words: building point cloud, moving-least square, normal estimation, feature extraction