Jump to main content
Professur Digital- und Schaltungstechnik
Computer Vision

Computer Vision

Our self-developed Framework for Computer Vision: XPCV
Please find demo videos of our algorithms at our Youtube channel
In order to use cameras (optical sensor) for measurements, a detailed knowledge of the underlying camera model is essential. Therefore, an accurate estimation of the camera model parameters and resource-efficient implementation is necessary.
With the help of contact-less, optical methods, vital parameters can be determined robustly and at the same time represent a significant increase in comfort for the persons concerned. The chair investigates the vital parameters heart rate, respiration rate, oxygen saturation and blood pressure in different wavelength ranges.
As part of its projects, the Chair of Digital Signal Processing and Circuit Technology is researching technical assistance systems for marker-less motion analysis. One field of application is the monitoring of exercises in medical training therapy.
As part of the project AUXILIA and other industry sponsored projects in the field autonomous driving, the professorship is actively researching in the area of object detection using deep learning.
Person detection algorithms can be used to determine the number of people in a room and their respective three-dimensional position and extent. Using this location data, long-term statistics about the location and duration of a person can be created. In addition to a person's position, their pose (standing, sitting, lying) is recognized by machine learning methods. The inclusion of posture information in long-term statistics allows more accurate conclusions about a person's mobility.
Various methods of stereo correspondence analysis are investigated and developed at the Chair of Digital Signal Processing and Circuit Technology. The aim of the development is to solve the practical challenges.
For modern computer vision applications, neural networks are often the key to success. Such systems are trained with huge amounts of data with the aim to learn different variations of objects. Depending on the domain, it can happen that the required training quantities are not available and already existing data sets cannot be adapted. One approach to counteract this problem is the generation of synthetic data.
The research area of ​​the professorship DST is not limited to the visible wavelength range (380 - 780 nm). Also the middle infrared range between 7,5 - 13 μm wavelength is subject of the research. The thermal radiation is detected with a thermal imaging camera. This process is called thermography. Since the representation of objects and persons differs considerably in this wavelength spectrum, image processing algorithms must also be rethought.
Stereo vision based sensor systems are a core technology to acquire RGB-D data for measurements of a 3D space in the context of different applications. This includes, for example, surveillance tasks in the interior, where a number of sensors work as a network to cover the complete area.