文摘
Modern computer vision and coordinate metrology systems provide an ever-increasing flow of information from the physical world we live into the virtual world inside computer systems. A wide range of applications including manufacturing inspection, planetary rovers, and computer-assisted surgery has emerged with the need to utilize this flow of information. Often the coordinate system of the vision sensor or metrology device has a different coordinate frame from that of the existing objects in the virtual world. To rectify these differences, a process called registration must be applied. The particular registration problem addressed in this work involves the relocating of Cartesian coordinate point sets such that they align with a corresponding parametric surface. The new method proposed in this dissertation is based on more traditional numerical optimization techniques rather than the Iterative Closest Point methods that are popular in this field. Past research has overlooked these traditional techniques due to the lack of analytic derivative information. This work provides a clear mathematical proof for these derivatives and uses them to implement a new registration technique based on Newton minimization strategies, yielding a registration procedure with distinct advantages in terms of accuracy, speed, and robustness.