Interactive Rapid Prototyping based on Computer Graphics for the Image Acquisition in Automated Visual Inspection
Final Report Abstract
In this project, we worked on the interesting problem of automatically proposing optimized setup constellations for optical measurement setups and, as an application, the laser triangulation inspection of a complex cylinder head object. One one hand, this project included development and research in the field of computer graphics for sensor-realistic simulation of the images in an optical inspection setup. We studied image simulations at several levels of fidelity, ranging from simulating the scene geometry only and the sensor coverage, to including further surface details such as measured surface reflectance for different materials on the cylinder head and measured surface roughness profiles, as well as simulating wave-optics effects such as speckle formation. We also took the specification of the camera sensor into account by means of the EMVA specification, which models the transformation of accumulated photons on the sensor to the final digital value. All this is combined to a detailed physically-based simulation of an optical imaging system in different steps of the light transport. In addition, real world experiments for verification of simulated images and comparison with real images have been also part of the project. Another focus of this project has been the study of error propagation in automated visual inspection systems, and a probabilistic perspective to surface reconstruction as a regression problem. These uncertainty estimations, together with the visibility information acquired from fast rasterization-based simulations have been subject to optimizations. We defined and studied the “inspection planning” problem as finding the minimum number of acquisitions that successfully inspect the product for the given inspection criteria and proposed solutions to solving it using a greedy approach and the particle swarm optimization. Planning results have been compared with the state of the art and the results demonstrated an improvement over the previous methods. All the related functionalities regarding the simulations and the automatic optimizations have been integrated into an expert application, which can be used for visualizations, expert validations, quick parameter tuning, and fast automatic optimizations. We believe that the developed methods, or similar approaches for other AVI systems other than laser triangulation, can significantly contribute to the design process of AVI systems and assist domain experts in rapid prototyping of such systems. This can be especially beneficial to the ever-evolving production lines with demanding inspection requirements.
Publications
-
(2018) Sensor-realistic simulation of images in diffraction-limited imaging systems / Sensorrealistische Simulation von Bildern in beugungsbegrenzten Abbildungssystemen. tm - Technisches Messen 85 (s1) s95-s102
Mohammadikaji, Mahsa; Bergmann, Stephan; Burke, Jan; Beyerer, Jürgen; Dachsbacher, Carsten
-
Surface inspection planning for laser line scanners. In Proceedings of the 2015 Joint Workshop of Fraunhofer IOSB and Institute for Anthropomatics, Vision and Fusion Laboratory (2015), Beyerer J., Pak A., (Eds.), Karlsruher Schriften zur Anthropomatik, KIT Scientific Publishing
Mohammadikaji M.
-
A Framework for Uncertainty Propagation in 3D Shape Measurement using Laser Triangulation. 2016 IEEE International Instrumentation and Measurement Technology Conference (2016), 6–11
Mohammadikaji M., Bergmann S., Irgenfried S., Beyerer J., Dachsbacher C., Wörn H.
-
A phenomenological approach to integrating gaussian beam properties and speckle into a physicallybased renderer. In Vision, Modeling Visualization (2016), Hullin M., Stamminger M., Weinkauf T., (Eds.), The Eurographics Association
Bergmann S., Mohammadikaji M., Irgenfried S., Wörn H., Beyerer J., Dachsbacher C.
-
Performance characterization in automated optical inspection using cad models and graphical simulations. In XXX. Messtechnisches Symposium. De Gruyter, 2016
Mohammadikaji M., Bergmann S., Irgenfried S., Beyerer J., Dachsbacher C., Wörn H.
-
Towards surface inference in industrial inspection. In Proceedings of the 2016 Joint Workshop of Fraunhofer IOSB and Institute for Anthropomatics, Vision and Fusion Laboratory (2016), Beyerer J., Pak A., (Eds.), Karlsruher Schriften zur Anthropomatik, KIT Scientific Publishing
Mohammadikaji M.
-
A versatile hardware and software toolset for computer aided inspection planning of machine vision applications. In International Conference on Information Systems Architecture and Technology (2017), pp. 326–335
Irgenfried S., Wörn H., Bergmann S., Mohammadikaji M., Beyerer J., Dachsbacher C.
-
Automatic planning for optimizing the surface coverage in industrial inspection. In Proceedings of the 2017 Joint Workshop of Fraunhofer IOSB and Institute for Anthropomatics, Vision and Fusion Laboratory (2017), Beyerer J., Pak A., (Eds.), Karlsruher Schriften zur Anthropomatik, KIT Scientific Publishing
Mohammadikaji M.
-
Cad-basierter Workflow für den semi-automatischen Entwurf optischer Inspektionssysteme. at - Automatisierungstechnik 65, 6 (2017)
Irgenfried S., Wörn H., Bergmann S., Mohammadikaji M., Beyerer J., Dachsbacher C.
-
Image formation simulation for computer-aided inspection planning of machine vision systems. In Automated Visual Inspection and Machine Vision II (2017), vol. 10334
Irgenfried S., Bergmann S., Mohammadikaji M., Beyerer J., Dachsbacher C., Wörn H.
-
Probabilistic surface inference for industrial inspection planning. In Applications of Computer Vision (WACV), 2017 IEEE Winter Conference on (2017), pp. 1008–1016
Mohammadikaji M., Bergmann S., Irgenfried S., Beyerer J., Dachsbacher C., Wörn H.
-
Inspection planning for optimized coverage of geometrically complex surfaces. In proceedings of 2018 IEEE International Workshop on Metrology for Industry 4.0 and IoT (2018), pp. 52–57
Mohammadikaji M., Bergmann S., Irgenfried S., Beyerer J., Dachsbacher C., Wörn H.
