Fraunhofer Institute for Physical Measurement Techniques IPM
Fraunhofer Institute for Physical Measurement Techniques IPM
Big Data: Messdaten-Auswertung
© Outflow_Designs / Shutterstock

Processing and efficiently evaluating measurement data in real time

Very rapid processing of measurement data is necessary if this data is to be used productively – for example, to control processes. At the same time, efficient methods for data evaluation are required in order to extract a maximum of information from a constantly growing amount of measurement data.

© Outflow_Designs / Shutterstock

Rapid evaluation of measurement data

State-of-the-art measuring devices and cameras generate ever increasing amounts of highly resolved data. However, tools for an efficient evaluation of measurement data are missing, so that their full potential is often not fully exploited. Companies with enormous expertise in hardware often break new ground in software and come up against their limits. In many cases, starting from scratch and acquiring own software development capacities turns out to be uneconomical.

Fast and efficient evaluation of measurement data is becoming increasingly important. Computing power that was inconceivable just a few years ago, is now widely available and provides the basis for this dynamic development. Against this backdrop, we have been developing tailor-made software solutions in our projects for many years in addition to measurement technology hardware. Among them are:

  • Data fusion of camera, 3D and other sensor data in real time
  • Analysis of large amounts of data by means of intelligent algorithms, e.g. AI based algorithms based on machine learning
  • Real-time data processing with very high bandwidths  in the production cycle

Which hardware is the most suitable? Which software tools can be used to extract the relevant information from the measurement data as efficiently as possible? With more than 40 years of experience in the development of optical measurement systems, we master the hardware in all its aspects. At the same time, we know the right strategies for merging multiple data streams. We will work with you to find the right tools for your specific measurement task and develop solutions to generate maximum benefit from your measurement data.

Some examples

 

© Fraunhofer IPM

Matching up to 10 million IDs per second

For 100 percent quality monitoring, camera systems record the microstructure of the component’s surface in the production cycle with high resolution. No additional markers have to be affixed to the product. From the recorded image, an ID is extracted that uniquely identifies this component. To unambiguously identify the components in the production cycle, up to 10 million IDs per second are compared.

 

 

 

 

© Fraunhofer IPM

Extracting 250 million 3D points per second from 25 Gbit/s of camera data

Digital holography enables high-precision 3D measurements in the production line. The technology allows components to be measured with micrometer accuracy over the entire surface. In this process, up to 25 Gbit/s of camera raw data is processed on graphic processors in real time to more than 250 million 3D points/s. This is how we detect even tiny defects.

 

 

© Fraunhofer IPM

3D reconstruction and data interpretation on a mobile device

For a digital documentation of underground infrastructure, we record georeferenced 3D data using mobile devices. The data streams are processed on the device in real time to create a 3D reconstruction. Image content is directly extracted using artificial intelligence. The result is a highly accurate yet lean digital 2D map with meta-information.

 

© Fraunhofer IPM

Fusing heterogeneous data to coherent data streams

On our customized measurement vehicles for the acquisition of environmental data, we merge the data from at least 4 high-resolution cameras, 2 panoramic cameras, several laser scanners and positioning and orientation systems into a coherent data set. The data volume is usually in the range of several gigabytes per kilometer.

 

© Holger Kock/Fraunhofer IPM

Simultaneously evaluating data from 27 cameras and 9 fluorescence cameras

During component testing in free fall, surface defects and geometric errors are detected. The components are dropped through a hollow sphere at one second intervals and recorded from different perspectives with the help of cameras. The measurement data from up to 27 cameras and 9 fluorescence cameras are evaluated simultaneously.

 

 

© Fraunhofer IPM

2 million 3D measurements – the fastest laser scanner worldwide

Our laser scanners capture up to 2 million 3D measuring points per second using a single measuring beam. This makes them ideal for use on fast-moving platforms such as trains, measurement cars or drones. Alternative systems achieve similarly high values only by parallelizing several scanners.

 

 

© Fraunhofer IPM

Evaluating 6 gigabyte of data per second

In the production of strip products, our camera-based systems control surfaces completely and continuously during production. This process generates up to 6 GBit/s of data, which are evaluated in real time.

 

© Fraunhofer IPM

Scalable analysis of terabytes of data

We extract targeted information from large amounts of data: Using AI, we fuse and analyze several terabytes of heterogeneous data (2D images and 3D point clouds) in a flexible and scalable cloud solution.

 

© /graphics Fraunhofer IPM

100 percent inline oil film measurement on metal sheets

The F-Scanner for oiling measurement in the production process consists of four laser scanners. The system enables 100 percent oil film control on large and fast moving sheets before forming. (4 ADC cards / 8 highly sensitive and fast detectors (photomultiplier tubes, PMT) / 1.6 MS/s)