Instrumentation for Estimating Surface Radiometry

Automating visual quality assurance of product appearance is a very hot topic in industry as it is still primarily a manual task only carried out by humans. With a strong focus on digitization in Industry 4.0, it is also highly beneficial for companies to be able to
include product appearance in their digital twin. In order to automate the visual quality assurance, we need to be able to model, measure, and compare appearance. Our focus is combining computer vision and computer graphics to develop the instrumentation and methodology needed to achieve the goal of automating appearance assessment. This thesis is focused around industrial applications in collaboration with industrial partners through an organization named Manufacturing Academy of Denmark (MADE).
Specifically, we contribute by developing three robotic setups that utilize cameras and controllable light sources to obtain surface information from radiometric measurements, ABB Setup, UR5 Setup and UR3 Setup. These instruments has been key components
in the contributions presented in this thesis, although they span a wide range of topics. One contribution is developing methods for measuring the contrast in engineered microsurfaces with the UR5 Setup. In relation to that, we developed a suitable BRDF model
for simulating these structures. By utilizing the high repeatability of the ABB Setup, we created a digitization pipeline that inputs data from multiple modalities and create a digital scene, which can be rendered and compared pixel by pixel with a photograph.
This allow for direct comparison of the digital and physical twin as images. We developed practical methods that can help introduce appearance in the digital twin such that it can be compared with the physical twin for Visual Quality Assurance.