Application of weighted total least squares based on intensity value in plane fitting of point clouds

Abstract

Aiming at the condition that each point in point clouds has different positional accuracy, a new method for fitting plane to point clouds is proposed. The method is based on weighted total least squares (WTLS). According to the effect of intensity of each point on positional accuracy and the structure of coefficient matrix, the weight matrix was defined. In order to verify the applicability of the proposed method, analytical experiments based on instant data of different materials were done, and comparisons between the method and traditional methods, such as least squares (LS) method and total least squares (TLS) method were implemented. The results show that the method is suitable for fitting plane to point clouds with good homogeneity, and more accurate parameters solution can be obtained as compared with TLS and LS. As for point clouds with poor homogeneity, the proposed method is unsuitable for the same operational use, while TLS is the most effective method. Keywords—point clouds, weighted total least-squares, intensity value, weight matrix, plane fitting