„The video evaluation systems“ department is divided into three research groups, which are briefly presented below.
Video-assisted security and assistance systems
„The Video-assisted Security and Assistance Systems“ group has been developing video exploitation solutions within distributed sensor systems for several years. These solutions have been developed in collaboration with security authorities and organizations to ensure that they meet high standards in terms of functionality, performance, ergonomics, and data privacy. The solutions cover a wide range of applications, including real-time video surveillance solutions such as the recognition of person activities, privacy-friendly person recognition, fine-grained auto-classification, and crowd monitoring. The research group also focuses on developing security-related image and video exploitation methods for analyzing large amounts of visual data, with the aim of detecting elements such as violence and weapons.
Group manager: Dr.-Ing. Jürgen Metzler
Video system technology
„The Video system technology“ group researches the software technology tools required for the rapid integration of current video evaluation methods into applications. The focus is on a high degree of system maturity and high system performance. The video evaluation system ABUL serves as the primary integration platform. The spectrum of integrated methods ranges from real-time video processing to deep learning algorithms for detection and classification. The system features multistream HD video processing, combining video data with geospatial information and other metadata. A video database that can be expanded in real time ensures efficient management of large volumes of data and enables geographic searching of the dataset. The system has been developed over many years in cooperation with video evaluators and clients with regard to functionality, performance and ergonomics.
Group manager: Dipl.-Inform. Jochen Ring
Image based reconnaissance
In the research group „image based reconnaissance“, automatic evaluation methods for images and image sequences are developed in the context of geo-information. The methods serve the discovery, recognition, identification and analysis of the depicted objects and scenarios. Here, methods of image processing are employed: e.g. Super-resolution, change detection, object tracking, pattern recognition, classification, deep learning / CNNs, geo-registration. The sensor system under consideration ranges from IR and video sensors via WAMI (Wide Area Motion Imagery) to SAR image sequences. Application areas are airborne and maritime applications for video analysis. For the application of these methods, the video evaluation system ABUL is developed in the department.
Group manager: Dipl.-Inform. Norbert Heinze