Abstract:In view of the misclassification caused by the traditional block-index curved filtering algorithm using fixed threshold, an improved block-index curved filtering method with adaptive threshold is proposed. Firstly, Gaussian filter and K-dimensionl tree filter are implemented to eliminate the abnormal point cloud. Then, to reduce the filtering effect of a extreme large seed region, the point cloud is blocked step by step with the grid method, and the seed points in the block area are automatically obtained by a way of surface fitting. An adaptive filtering threshold model, thus, which considering two factors including the size and maximum height difference of the block area, is established. The filtering performance is compared between the traditional method and improved method with three different sets of data, respectively, the results show that the proposed method can not only solve the problems caused by manual selection of seed points, but also effectively reduce the two kinds of errors, which verifies the reliability of the improved algorithm.