Machine vision is one of the fastest growing sectors in the German and European mechanical engineering industry. According to the latest VDMA figures, companies in this sector generated 18% more turnover in Germany in 2017 than in the previous year. The increase was also significantly higher than the average increase in turnover in the mechanical engineering sector across Europe. One growth segment is hyperspectral imaging. This relatively young discipline, also known as hyperspectral imaging (HSI), is said to have enormous potential for the future.

Markus Burgstaller, CEO of Perception Park, describes the main difference to traditional vision systems as follows: "Compared to conventional image processing systems, hyperspectral systems offer a spectrum per object pixel instead of a monochrome or color value. Depending on the wavelength range and spectroscopic processing, high-precision color coordinates, chemical material properties and even layer thickness information can be derived from the spectral data. The output information of such a camera has a significantly higher degree of complexity, but also enables a much higher diversity and selectivity with regard to solvable applications." "While conventional RGB cameras only image the colors red, green and blue, hyperspectral cameras allow the differentiation of more than a hundred colors," confirms Dr. Jan Makowski, Managing Director of Luxflux. "This type of high-precision colour measurement can be used to examine the properties of substances and make chemistry visible."

Diverse fields of application

Tim Huylebrouck, Product Manager at Stemmer, hints at the possible applications behind this method: "Objects that supposedly look the same can - when excited with a broadband illumination source - reflect completely different light spectra due to their chemical properties. These can then be distinguished from one another using hyperspectral systems. No other image processing solution can do this in this form."

Dr. Georg Meissner, Managing Director of Specim, cites an exemplary application: "Zenrobotics is a leading global supplier of robotic waste sorting systems. As the building rubble that is sorted in such plants often contains hazardous materials such as asbestos, an important requirement is to identify such materials safely and reliably. Zenrobotics therefore relies on Specim's hyperspectral cameras, which offer the necessary detection reliability, sensitivity and speed for this task."

Gion-Pitschen Gross, Product Manager at Allied Vision, also sees the recycling and sorting of plastics as an important field of application for hyperspectral systems: "HSI enables the automatic separation of plastic parts, for example made of polyethylene and polypropylene, which can be identified and separated based on their chemical composition. In addition to the existing color sorting, materials can be differentiated according to their molecular composition. This significantly improves the quality of the sorting process." According to Gross, the inspection of food also holds great potential for HSI technology: "Meat, fat and bones have differences in their molecular properties that can be clearly recognized in an HSI image. This also applies to other materials that hardly show any differences in the real image, such as sugar, salt and citric acid, which appear almost identical." It is also difficult for cameras in the visible spectral range to recognize physical changes to objects. This plays a major role in the food sector, for example, when fruit or vegetables are to be inspected for their degree of ripeness or possible mold infestation. This is where HSI systems offer suitable solutions, which Allied Vision addresses with its hyperspectral cameras from the Goldeye series, among others.

Daniel Hofmann, CEO of Spanish Photonfocus subsidiary Solpi, expects a breakthrough in HSI applications with mobile carrier systems such as precision farming with the help of UAVs (Unmanned Aerial Vehicles) in the future: "Camera systems can be mounted on a drone to enable photogrammetry or inspection applications. Such a camera system can consist of several hyperspectral cameras, a GPS system, an embedded computer and much more. The captured images are provided with precise GPS data to simplify subsequent image processing." Solpi offers a camera system that allows the use of multiple hyperspectral cameras in a stand-alone grabbing solution.

Challenges for HSI image processing

Despite these and other promising application examples, hyperspectral imaging is currently still one of the exotic disciplines in image processing. One reason for this is a number of challenges that still need to be overcome before the technology can be used across the board. The still relatively high price of hyperspectral technology is one of the main barriers to entry. In addition, the entire technology is not easy to understand and often requires in-depth specialist knowledge in the field of spectroscopy.