Reconstruction of Moving Surfaces of Revolution from Sparse 3-D Measurements using a Stereo Camera and Structured Light

The aim of this thesis is the development and analysis of an algorithmic framework for the reconstruction of a parametric model for a moving surface of revolution from a sequence of sparse 3-D point clouds. A new measurement device with a large field of view that allows for acquisition of three-dimensional data in challenging environments is utilized. During the measurement process, the observed object may be subject to motion which can be described in terms of an analytical model. The proposed method is developed and analyzed, along with an application for the surface reconstruction of a wheel. It is shown that the precision of the coarse surface model independently fitted to each measurement can be significantly improved by fitting a global model to all measurements of the sequence simultaneously. The global model also takes into account the object's motion. The three-dimensional point clouds are acquired by an optical device which consists of a stereo camera and an illumination unit projecting a dot pattern. A rather high density of surface points within the camera's field of view is established by means of multiple laser projectors. Through an elaborate calibration procedure of the stereo camera and the projector, and by utilizing the trifocal epipolar constraints of the measurement device, a high accuracy in the three-dimensional point cloud is achieved.