Multi sensor industrial metrology for high precision measurements and 3D modelling

点击翻译按钮获取中文摘要

The primary objective of this research is the development of algorithms and procedures for the extraction of high precision 3D data of objects in an industrial environment. As the project evolves, research efforts will focus on the semi-automation of the feature extraction process from digital sensor sequences taken from various viewing locations. This would include image locations at various heights, distances and angles with respect to the object being imaged as well as the use of varying imaging modalities. The pursuit of this line of research was inspired by the need for precise three-dimensional data for reconstruction, modelling and geospatial information systems. As some of our infrastructure and industrial facilities begin to show signs of aging or increased demand requires upgrading, there is a need to replace, repair and modify existing structures and systems in order to comply with maintenance schedules and evolving demands. In many cases, original plans are unavailable or non-existent and, at best, are simply 2D plans for a three dimensional structure. The complexity of the structure can vary over time with incremental additions and modifications that are often unrecorded. During the retrofit or upgrade phase, 3D plans and graphics are essential for achieving an efficient and precise modification. Many of these structures and facilities involve the use of components that are comprised of basic geometric forms that can be modelled using geometric primitives. Since my move to the Department of Civil Engineering at Ryerson University, we have established a new digital metrology laboratory that is equipped with several high-end computers, digital cameras and access to high resolution terrestrial Lidar scanners. Through the use of these cameras and Lidar systems, the information content of high-resolution data sets can be exploited to capture high-precision 3D information of structures. This research will address the problems associated with data fusion from a multi-sensor acquisition environment.