Research on point cloud registration
YU Ming-x · Technology and Economy in Areas of Communications · 2013
Registration for point clouds is so important in 3D object modeling that it directly influence the final synthesis result and model precision.The main problem in point cloud registration methods is that the demanding original position registration.The registration for point clouds are divided into two phases.The first phase is coarse registration,by using man-machine interactive and saving time.The second stage is the registration,based on the least-square principle,using the indirect adjustment ideas,through the nearest point iterative algorithm for fine registration and using the target concentration points coordinates into the target concentration between registration accuracy of error as an index in the evaluation.Experimental results show the registration accuracy of this algorithm reaches to a satisfied level.So this experiment method is effective and feasible.