Extrinsic Calibration And Accury Assessment Based On Plane Features Of Vehicle Mobile Mapping System

A vehicle mobile mapping system(VMMS),as a new geographic information device of surveying and mapping,with its advantage of high-efficiency,high-precision and strong real-time in data collection,has become increasingly important in the production of mapping geographic information.In order to determine the relative positional relationship of sensors accurately and improve the accuracy of the vehicle mobile mapping system,calibration must be performed before the system operates.In this paper,at first we summary and analysis the affecting factors on the system accuracy,according to the equations for laser scanning point positioning.The method of external orientation element calibration and accuracy evaluation for the VMMS based on planar features is proposed.The main works of this paper are as follows:1.According to the coordinate transformation relationship of sensors in the VMMS,the geometric location equation is derived and the error sources are comprehensively analyzed,which include observation,placement,attitude and comprehensive errors.In the placement errors,the angle error is obvious and is proportional to the distance,while the translation error is smaller relatively and is a fixed value.2.On the basis of the analysis of the influence law of the angular error,a method for extrinsic calibration of the VMMS based on planar features is proposed.There are two main calibration methods,the one is the calibration method which is based on plane control and constraints,and the other one is a more generalized method of splicing point clouds based on planar features.And then these methods are verified by experiments.3.A method for accuracy assessment of the VMMS based on reference plane is proposed.Inspired by the derivation of the classical Ferrero formula,the accuracy of the system is evaluated by reference to the plane point cloud.Due to the feature points are difficult to be extracted,the traditional precision evaluation method has some limitations in evaluation result,behaving in unreliable,inaccurate and repetitive.While the method proposed in this thesis can overcome this limitations.The feasibility of the method is demonstrated by reference to the traditional point cloud accuracy assessment method.