intelligent AI solutions for hyperspectral imaging
intelligent AI solutions in hyperspectral imaging Heinrich Grüger is a physicist, holds a doctorate in materials science and works in the OASYS project in the hyperspectral imaging cluster as lead project manager at the IPMS in Dresden and Cottbus. He has been developing sensors and sensor systems at the Fraunhofer IPMS (Institute for Photonic Microsystems) for over 25 years and has a personal passion for optical spectroscopy, particularly in the near-infrared (NIR) spectral range. This is the range that can no longer be seen with the normal eye, but which is used in everyday life for information transmission, e.g. in TV remote controls. “I always find it super exciting when our microsystems technology, i.e. the smallest objects and nanometer-thin layers, meet the real world ” Scientist Heinrich Grüger explaining his work at the Long Night of Science in Dresden The well-researched method of NIR spectroscopy can be used to carry out application-oriented real-time measurements in areas such as food quality, health care or textile recycling. >> Every stain, every fabric, every object has a specific spectral fingerprint << So what is so special about hyperspectral imaging? Heinrich Grüger explains it to me using an example. If you take a photo of any object with a cell phone, e.g. a T-shirt, it is reproduced on the cell phone display in digital pixels, using the associated color information RGB (red, green, blue). With hyperspectral imaging, on the other hand, the chemical composition of each pixel is also obtained (chemical sensing). This is made possible by near-infrared spectroscopy, in which hundreds of wavelengths are measured simultaneously (so-called spectral channels). The information read out from the wavelengths can now be assigned to each pixel, thus indicating the composition. Each chemical compound has a specific spectral fingerprint. This makes it possible to measure whether a T-shirt is made of cotton or polyester fibers and the corresponding properties can be determined (washable at 40 or 60 degrees). Quite useful if the label on the shirt is missing or no longer legible. Stains on items of clothing can also be identified using this method. >> New fields of application in recycling are opened up by reducing time and costs << With the OASYS project, Heinrich Grüger is now working on an intelligent AI-supported solution for hyperspectral imaging in order to reduce the amount of data collected to a minimum. The composition of a T-shirt does not have to be recorded everywhere. A few measuring points are sufficient, which saves time and drastically reduces the associated costs. >> This new, smarter technology makes it easy to revolutionize many existing and future recycling processes << In the sub-project “A1 intelligent ultra-compact hyperspectral camera”, hardware and software for the compact near-infrared spectrometer are developed as well as the artificial intelligence with corresponding decision algorithms for spectral analysis. The relevant information is extracted from all measurement data, evaluated internally and displayed in a user-friendly way. In the recycling process, this technology is very useful for deciding whether the textile gets a second chance for further use (reuse), whether it belongs in recycling – where the fibers are recovered – or whether it is so used up that it can ultimately only be disposed of or incinerated. >> Better decisions are made on the basis of measured knowledge << With other objects and decision-making processes, for example in fruit sorting plants, algorithms can be used to decide in real time whether the freshly harvested apple is suitable for display in the supermarket or whether it would perhaps be better delivered to the cidery for juice production. Internal bruises in fruit and vegetables can be invisible to the naked eye from the outside, but with harmless near-infrared spectroscopy they become clearly visible. >> The research and development opportunities in the OASYS project fascinate me because the topics are so closely linked to everyday life << Fruit sorting is a particularly good field of application where you can make a contribution. In a short period of 5-8 years, the technology developed will move from the laboratory into everyday life. New technical solutions ultimately enable a targeted contribution to socially relevant topics, such as reducing food waste, avoiding waste or minimizing CO2 emissions. Heinrich Grüger has been enthusiastic about this at Fraunhofer right from the start. He can work scientifically here and has the opportunity to implement his own ideas. He is delighted to experience his own technological developments in real-life applications. Personally, he is most interested in how the individual sensors developed and, in future, the combination of different sensors with the help of AI can create added value in order to open up new fields of application and thus solve everyday or special problems. >> I would like to explain the topic of MEMS in a child-friendly way on “Sendung mit der Maus” << Heinrich Grüger has been involved in promoting young talent for a very long time and very intensively. His intention is to get children and young people interested in science and technology as early as possible. One of his dreams is to explain the topic of MEMS (micro-electro-mechanical systems) in a child-friendly way on the “Sendung mit der Maus” TV show, which he really likes because of the simple way it conveys knowledge, in order to get others interested in the topic of microsystems technology. Application examples Apple sorting Easier quality control with intelligent portable spectrometer: Bruises become visible without damaging the fruit All Posts eng Nicht kategorisiert project A1 hyperspectral ultra compact AI camera read more Background & expertise OASYS Cluster A: hyperspectral imaging By researching promising sensory components, the OASYS project is creating the basis for new technologies in a variety of innovative fields of application that drive the development of better processes, e.g. in the medical field of gentle (non-invasive) examination methods, machine-aided industrial production, optimized process technology, intelligent recycling, modern agricultural production, smart mobility applications and consumer electronics. Read more All Work /Interview Intelligente KI Lösung für die hyperspektrale Bildaufnahme Use cases All Work…
