Machine Vision (MV) and the related, but distinct, field of Computer Vision (CV) refer to the development of technology and methods that allow the automatic inspection and analysis of images for industrial production purposes (MV), or the duplication of the ability of humans to visually describe his/her environment (CV). The research group Machine and Computer Vision Applications uses various vision technologies for applications in industrial inspection and visual perception-based robotics. 2D- and multi-dimensional measurements are taken with the aid of imaging sensors, an information processing system to construct the necessary algorithms, and an integrating mechatronic superstructure.
The advantage of the machine lies in the versatility of its sensors, reproducibility during task performance, as well as its ability to process large amounts of data in real-time. Spectral diversification and the possibility to observe not only the outer surface of an object, but also the inside, are properties that represent major advantages and highlight the versatility of machine-based sensing .
For decades, JOANNEUM RESEARCH has built up sound technological expertise in this research area, which primarily explains the rate of success to date. Many of the developed application-oriented technologies are the result of the combined efforts by enthusiastic team members to understand customer needs and challenges. These efforts often result in the formation of an emotional bond to the final product (tunnel surveying, sports, space research, food industry, etc.).
The number of algorithms constructed in the areas of image processing, pattern recognition and classification within the context industrial inspection, robotics, metrology and human-centered image analyses is constantly increasing. The improved ability to miniaturize sensors allows a range of cognitive, affective and physiological data for use in behavioral studies to be collected via multi-sensoral human-machine interfaces and interpreted using innovative methods of image processing and machine learning. Our experts focus, in particular, on topics of medical image processing and biometrics and implement solutions in accordance with the Medical Devices Act. A larger range of research and applications than ever before can be covered with reference to industrial inspections in 2D (i.e., directly from images) and 3D (by combining multiple images or using specialized 3D sensors such as laser scanners). This is made possible by the use of new camera technology, data transmission technology and multi-GPU parallelization, as well as new illumination technologies (LED light sources). Robotics, on Earth as well as in space, represents a profitable niche in which mainly 3D vision techniques are used and that, through the integration of photogrammetry expertise during the recent year, has ensured the accessibility and sustainability of the research group.
Within the framework of the research projects, a broad network that involves national and international research partners was built. Additionally, a versatile laboratory infrastructure was created. This infrastructure consists of an image processing laboratory, an assembly line, a mobile robotic platform and adaptable tools for 2D- and 3D-shape measurement. Therefore, feasibility studies can be carried out efficiently and rapidly, even down to the level of custom-made products for use in industrial environments. We can offer support for research and product development, as is the case for medical products, which is ISO13485 compliant.