Abstract: In order to improve the precision of profiles reconstruction,new algorithms to process point data based on sliced method have been proposed.At first,a refinement algorithm has been adopted to refine the sliced point data,which applies Similar Moving Least-Squares tracing method.The whole computing process does not change the local coordinates of the measured points,and the step size is controlled by the cloud density.A double link-list method is used to sort the refined points.To extract feature points,by combining angular deviation and chord deviation,a new redundant point data processing and feature extraction method has been put forward.The proposed method can properly group the points and generate proper profiles.