A Framework for Position Uncertainty of Unorganised Three-dimensional Point Clouds from Near-monostatic Laser Scanners Using Covariance Analysis

Position uncertainty is one of the most important quantities of an unorganised three-dimensional point clouds since it provides the confidence level of any parametric estimation such as surface normal vector estimation and the registration of point clouds. We present an explicit form of position uncertainty based on the covariance analysis of a point. In addition, an explicit form of the variance of an estimated surface normal vector and an algorithm to evaluate an optimal size of the neighbourhood of a point which minimises the variance of the estimated normal vector are presented.