A new approach to stitching optical metrology data

The next generation of optical instruments, including telescopes and imaging apparatus, will generate an increased requirement for larger and more complex optical forms. A major limiting factor for the production of such optical components is the metrology: how do we measure such parts and with respect to what reference datum This metrology can be thought of as part of a complete cycle in the production of optical components and it is currently the most challenging aspect of production. This thesis investigates a new and complete approach to stitching optical metrology data to extend the effective aperture or, in future, the dynamic range of optical metrology instruments. A practical approach is used to build up a complete process for stitching on piano and spherical parts. The work forms a basis upon which a stitching system for aspheres might be developed in the future, which is inherently more complicated. Beginning with a historical perspective and a review of optical polishing and metrology, the work presented relates the commercially available metrology instruments to the stitching process developed. The stitching is then performed by a numerical optimization routine that seeks to join together overlapping sub-aperture measurements by consideration of the aberrations introduced by the measurement scenario, and by the overlap areas between measurements. The stitching is part of a larger project, the PPARC Optical Manipulation and Metrology project, and was to benefit from new wavefront sensing technology developed by a project partner, and to be used for the sub-aperture measurement. Difficult mathematical problems meant that such a wavefront sensor was not avail able for this work and a work-around was therefore developed using commercial instruments. The techniques developed can be adapted to work on commercial ma chine platforms, and in partuicular, the OMAM NPL/UCL swing-arm profilometer described in chapter 5, or the computer controlled polishing machines as manufactured by the project's industrial sponsors,- Zeeko ltd. The novel process involving physical surface fiducialization is then developed for a standpoint of instrument particularities followed by the development of a numerical optimization to derive the best stitching. The software, developed in Matlab, is then tested on simulated and real data and shows much promise. The stitching accuracy is found to be dictated by the quality of the acquired data and the precision with which fiducials in a given measurement can be related to each other. Finally a summary is given for a future direction of this work and how it might be improved upon