“Big Picture”: measuring data giants
Digitising three-dimensional objects to high levels of precision produces image files in sizes that have never been seen before. Conventional digital image processing is no longer enough to handle these huge, highly complex volumes of data. The "Big Picture" research project is therefore breaking new ground in the storage, processing and analysis of this image data - now with the help of a mummy more than 1000 years old.
"The aim of the ‘Big Picture' project is to use various non-destructively measuring sensor systems to obtain information that will for example enable us to derive control and regulation measures for process monitoring", says Prof. Tomas Sauer, head of the "Knowledge-Based Image Processing" Fraunhofer research group and of the University of Passau's FORWISS Institute. The large volume of data in particular requires the research and development of image processing strategies and procedures using new, intelligent approaches taken from the world of machine learning.
The subject currently being tackled by the "Big Picture" researchers is a Peruvian mummy over a 1000 years old. "We used high-resolution computer tomography to generate a terabyte of data. To enable us to process it, my team developed a program that allows us to compress the data to 5% of its size without any appreciable loss of quality. The next step is to find interesting areas and define segmentation tasks - and ultimately we also want to use automated learning methods to teach the system to be able to identify individual image areas on its own", Tomas Sauer explains. One nice side effect of this is that high-quality image capturing also enabled researchers to establish new facts and historical information about the mummy, an exhibit in Stuttgart's Linden Museum.
The procedures developed here can in principle be applied to numerous different sectors: the automotive industry, for example, where sensor technology and image processing via computer tomography can be used for large, safety-related cast iron components, or where crash cars can be examined on a purely virtual basis. "Tomography enables production-integrated control of safety-related components such as cylinder heads. ‘Big Picture' is seeking to develop solutions that significantly improve this technology's applicability", Sauer explains.
|Project period||01.03.2017 - 29.02.2020|