With High-Speed-Microscopy, large-area objects can be acquired in a very short time on a microscopic scale. In industries such as semiconductors, electronics and biology, the demand for microscopic inspection methods is particularly high, as the increasing inspection of large surfaces requires an ever faster acquisition process.
In contrast to conventional stop-and-go operation, high-speed microscopy allows the object to move continuously during the recording process. This allows the sample to be digitized at very high frame rates - more than 100 fps depending on the camera. Synchronous flash illumination avoids motion blur. During the continuous scanning process, real-time hardware autofocus systems automatically adjust the focus so that the sample is sharply imaged at any point. The individual images are stitched together by an efficient stitching algorithm to form a high-resolution, large-area overall image.
Software for real-time data acquisition with RAM recording, which was developed at the Fraunhofer IPT specifically for high-speed microscopy, ensures that image preprocessing and stitching run without delay thanks to GPU processing. The images can be automatically evaluated using specially developed image processing algorithms. Current developments are aimed at the use of machine learning methods such as Deep Learning using Convolutional Neuronal Networks (CNN).
Although the process is already around 30 times faster than comparable stop-and-go methods, the Fraunhofer IPT is working on accelerating the recording process even further. To this end, in addition to constantly improving the software, new components are regularly integrated, such as flash illumination, sensor technology for the hardware autofocus system and camera technology.
- Development of high-throughput microscopy solutions with customized software
- Enabling existing microscopes for high-speed image acquisition
- Development and installation of high-performance short-pulse illumination systems
- Individual image data evaluation, classical and AI-supported
- Integration into automated systems