Enhanced-ICP Algorithm For Registration

With the rapid development of measurement technique and computer technology, reverse engineering is widely applied.3D point cloud registration algorithm is an important part of computer technology. We could scan an object from different viewpoints by 3D laser scannings, but if we want to output the whole model, we should register these range images through rotation and translation.This paper is mainly for pre-processing and registration of point cloud. The innovations and improvments of this article are mentioned as follows:The outlier detection filtering of scattered point cloud by classification are proposed to simplify the point clouds. Trough this algorithm, we can detect and divide massive point clouds into far outliers and near outliers, which is based on their distances to the main body of the points.The experimental results show that the completion of the steps above enable us to get clean point clouds,which makes the registration more robust.Enhanced particle swarm optimization combined with iterative closest point algorithm is given. ICP will fall into local optimal if the two sets of point cloud are extremely different in initial directions and positions. We constrain not only the distances between the point pairs but also the directions of their normal vectors.The experiments conducted shows that the accuracy and speed of the registration are both improved.Dual quaternion algorithm is proposed to optimize the spatial transformation.This method will coordinate transformation into the real and the former Ministry of two parts, such as by obtaining mathematics eigenvalues and eigenvectors obtained by the method and, respectively, can get the value of the rotation matrix and translation matrix, not like the singular value decomposition and unit quaternions obtained by rotating pan variable variable, the method reduces the error of the measurement data and cloud computing the amount of rotation T in our experiments brought by the ICP method is applied to the registration, the increase in the rate of registration and accuracy.