Yao Chunjing, Zhou Qihang a
In this paper, an automatic dam extraction algorithm which is called polynomial curve fitting is proposed. This method is based on the LiDAR points cloud filtering and the spatial geometric distribution feature analysis. The key ideology lies in the following steps: firstly, obtain reference vectors from water profile (airborne LiDAR)or vehicle trajectory file (mobile LiDAR); secondly, a section with certain width is made along the direction of dam and the current points in that section are projected to the two-dimensional space which is vertical to the direction of the dam; then, divide the projected points along the horizontal axis into several segments andsearch a partial lowest point in every small segment, after which the polynomial coefficients of two-dimensional can be solved with least square method; lastly, calculate the distance between the point in dam section and the fitting curve, the current point can be judged as dam point when that distance to the curve is smaller than the given threshold, otherwise is not. The experiment with the vehicle LiDAR data of JingJiang Dam shows that this method can effectively extract dam point and the accuracy can be better than 0.16 m.