Development of a fast and cost-effective computed tomography system for industrial environments by incorporating priors into the imaging workflow

• Autor(a) principal: PEREIRA, Luis Filipe Alves
• Data de Publicação: 2018
• Tipo de documento: Tese
• Idioma: eng
• Título da fonte: Repositório Institucional da UFPE
• Texto Completo: https://repositorio.ufpe.br/handle/123456789/29992
• topic: Ciência da computação; Tomografia computadorizada

Conventional X-ray radiography has been extensively used for inspection and quality assurance of industrial products. However, 2-D X-ray radiography cannot provide quantitative information within three dimensions about the scanned object. To obtain such depth information, X-ray Computed Tomography (CT) should be applied. Nevertheless, conventional CT systems (at which the X-ray source and detector rotates around the target object) are cost ineffective, inflexible, and suffer from long acquisition times. Therefore, the deployment of such technology is unfeasible for many industrial environments where high throughput is required as much as the best cost-benefit rate. The main goal of this research is to design a simple and cost-effective X-ray CT imaging system of high throughput for industrial environments. This system should comprises a single and static pair of X-ray source and detector for imaging objects passing on a conveyor belt. Such setup has been widely used with traditional radiographs for quality assurance in industrial environments; however, the large number of unknown projection views made such setup unfeasible for CT. Computer vision- and machine learning-based improvements are applied to incorporate prior knowledge about the scanned object into the CT imaging workflow as a way of compensating the lack of multiple X-ray sources or moving parts in both source and detector. More precisely, it is evaluated the use of priors related to the materials composition and also the outer object shape, as well as the use of Machine Learning techniques to apply priors automatically extracted from a training set of previous reconstructions. The trade-off between reconstruction quality and system’s throughput is exposed by linking the following measures: processing time, conveyor belt acceleration/deceleration, number of X-ray projections, reconstruction accuracy, and image resolution. It is also shown that one of the proposed methods can improve the system’s throughput in 21% while keeping the reconstruction accuracy over 90%. This research represents an advance in the state-of-the-art since it demonstrates that is possible to generate good quality reconstructions from projections acquired in an usual scanning setup where both X-ray source and detector are statically positioned.