Mathematical Models and Image Analysis Algorithms for Industry
The department develops mathematical models and image analysis algorithms and converts them into software suitable for industrial use, primarily in production. The application areas include in particular sophisticated surface inspection and analysis of microstructures. We have been developing and distributing software for 2D and 3D image analysis for over 15 years and develop both new methods and domain-specific machine learning algorithms.
In recent years, one focus in machine learning has been on image processing for production and industry. Methods such as »Deep Learning« require a high number of annotated data, for example of the defects to be found in a production plant. Now, however, in a well-functioning manufacturing plant, there are many images of defect-free products, but only a few of products with defects. We therefore often use hybrids of the »classical« parameterizable methods (filters, morphology, edge detectors) and machine learning. In addition to solutions for production, we also offer »typical« machine learning solutions for image processing. These are often projects in which very large amounts of image data are processed manually and this process is to be automated by software.
Another focus is microstructure analysis. The microstructure of modern materials significantly determines their macroscopic material properties. We develop algorithms for characterization and stochastic modeling of such microstructures based on image data, e.g. from CT, FIB-REM, SEM.