Fraunhofer IAIS is a global leader in the areas Large Scale Data Mining and Pattern Classification and has gathered years of experience in the evaluation of image and, in particular, hyperspectral data. To ensure correct allocation, Fraunhofer IAIS uses algorithms from the area of machine learning. When using machine learning - a subfield of computer science - an algorithm is trained within the framework of an initial test phase. In the case of blackValue®, this means that the developer shows the system various samples and also provides information about the respective material. In the second test phase, the developer shows the system further samples and the system must allocate the samples to the various categories on the basis of its training.
The programming of the algorithm does not involve a special programming language: the key to success lies in the researcher's in-depth knowledge of the methods used by programming languages that are commonly used today. This paves the way for the creation of efficient but lean algorithms that can achieve the required precision with a decision time of just one millisecond while at the same time reducing the burden on the hardware to a minimum to ensure that the software is also suitable for standard industrial computers.