Building depth images from scattered point cloud

Abstract

With the equation of plane and sphere, we fit them with Linear Least Squares. To cylinder datum fitting, firstly parameterize GQS equation of cylinder from seven parameters to five parameters, then using Local Paraboloid Construct method based on coordinate translation to get fitting initial values, finally evaluate results by Levenberg-Marquardt--a Nonlinear Linear Least Squares. Algorithm. However, initial values with Local Paraboloid Construct method are unstable. So to improve the precision of cylinder fitting ,a robust cylinder fitting method is put forward, which at first gets initial cylinder parameter values by Gauss Image, then fits cylinder by Nonlinear Least Squares for parameterized distance function. After getting reference datums, this paper proposes the methods of creating depth images from scattered point cloud and the specific steps with reference to different datums. Finally we choose some point cloud data of ancient building components from laser scanning data of Forbidden City in China as experiment data. Experiment results demonstrate the stability and high precision of the method of plane, cylinder and sphere fitting as well as the validity of depth images to represent point cloud of object.