Optimization of parameter settings by projection based quality parameters in industrial computed tomography

制造业中的新技术（如增材制造）能够生产对精度和高功能集成度有高要求的零件。这些部件通常包含无法用传统触觉坐标测量法访问的内部结构和隐藏特征。因此，计算机断层扫描（CT）变得越来越重要，因为它是唯一可以无损测量内部结构的方法。计算机断层扫描提供了分布在零件整个内外表面几何形状上的高密度测量点，使测量部件的三维体积模型成为可能。用于尺寸计量的计算机断层扫描是一种复杂的间接测量过程，其结果取决于各种影响因素。从测量数据的数据处理的单一投影开始，对于每个测量任务，必须相应地选择单独的参数集。测量的质量和可靠性（以测量不确定度表示）取决于硬件和软件以及用户设置的扫描参数。 扫描参数，如电流、管电压或曝光时间，会影响测量结果。特别是在微型计算机断层扫描，这种影响是显着的。然而，在大多数情况下，扫描参数的选择仍然是基于CT用户的经验，而由于复杂的测量过程，参数选择对测量结果的影响无法量化。因此，在所提出的项目中，开发了一种关于自动设置扫描参数的方法，其目的是减少特定于任务的测量不确定性。首先，在模型中评估并定量描述各个投影的图像质量参数对重建体积和测量偏差的影响。为了测量，使用显示出与真实的工业部件相似的测试体。因此，特别感兴趣的是关于几何特征的尺寸和形式的测量误差之间的差异。基于该模型，然后可以从该模型导出投影的所需图像质量，并且可以确定用于每个测量任务的单独的一组最佳图像质量参数。目的是开发和验证一种自动优化扫描参数的方法。

New technologies in manufacturing (such as additive manufacturing) enable the production of parts with high requirements towards accuracy as well as a high functional integration. These parts often contain inner structures and hidden features that cannot be accessed with conventional tactile coordinate metrology. Computed tomography (CT) thus is getting more and more important, because it is the only method that can measure inner structures non-destructively. Computed tomography offers a high density of measurement points spread over the whole inner and outer surface geometry of the part, making a three-dimensional volumetric model of the measured component possible. Computed tomography for dimensional metrology is a complex and indirect measurement procedure, whose results depend on a variety of influencing factors. Starting from the single projection towards data processing of the measured data, for each measuring task an individual set of parameters has to be chosen accordingly. The quality and reliability of the measurement, expressed in measurement uncertainty, depends on hard- and software as well as user-set scan parameters. Scan parameters, such as current, tube voltage or exposure time, can influence the measurement results. Especially in micro computed tomography this influence is significant. However, in most cases the selection of scan parameters is still based on the experience of the CT user, while due to the complex measurement process the influence of the parameter choice on the measurement result cannot be quantified. Thus, in the proposed project, a methodology regarding an automated setting of scan parameters is developed, which aims at a reduced task-specific measurement uncertainty. First, the influence of image quality parameters of the individual projections on the reconstructed volume and respectively on the measurement deviation is evaluated and quantitatively described in a model. For the measurements, test bodies that show similarity to real industrial components are used. Of special interest hereby is the difference between measurement error regarding dimensions and form of geometric features. Based on the model, the required image quality of the projection then can be derived from the model and an individual set of optimal image quality parameters for each measuring tasks can be determined. The aim is to develop and validate a methodology for the automated optimization of scan parameters.