Point cloud matching with a handheld Texel camera

This thesis investigates two major topics: one, to develop a small, handheld device that models the surface of a real scene and two, to develop an algorithm that is able to locate the relative shift of one surface to another overlapping surface from data taken from such device. The handheld device is used as a method of capturing data in a real scene without, the need of bulky equipment. The ability to locate the relative shift of one surface to another overlapping surface allows the device to be able to capture 3D (three-dimensional) models of scenes that do not fit into the field of view of the camera at its resolution.;Also included in this thesis is a method of calibrating the handheld device such that distortions viewed in the images are minimized and the ladar and visual images can be mapped to one another by a linear relationship. This is important so that the models formed are not subject to the non-ideal properties of the mechanical and optical pieces of the handheld device.;Finally, this thesis gives results for several scenes taken by the handheld camera as well as how well the algorithm matches several overlapping scenes taken from said device. These results show that the root mean squared error of the algorithm developed for this thesis for matching overlapping scenes together is at most about double that of the variance of the poses by themselves.