In order to improve detection quality of video systems, while at the same time reducing the false-alarm rate, we need brand new concepts, architectures and algorithms. The new, high-resolution cameras currently available on the market along with more complex algorithms do allow for an enhancement of quality levels, but they also require a higher computing performance of the systems. Novel hardware architectures, such as multi-core concepts, are therefore needed to push ahead with these new developments. For the further evolution of processors and the new breed of multi-core technologies, the creation of new software architectures and algorithms, which are able to meet unprecedented requirements, is also imperative.

Essential improvements include:

• higher performance
• higher resolution (with exponentially growing data rates)
• higher quality for detection and tracking
• increased algorithmic complexity for more robustness with less false alarm rate
• better usability

Furthermore, image processing algorithms must be effective also under difficult illumination conditions in both indoor as well as outdoor environments.

The Safety & Security Department has extensive expertise in optical 3D Stereo Vision that can be regarded a powerful application of multi sensor data fusion. The combination of two or more camera images taken at the same time, but from different viewpoints, allows constructing a 3D model of the scene. Research on this topic will continue and results will be applied in autonomous systems for obstacle and lane detection as well as navigation being part of multimodal sensor systems.

New concepts, architectures and algorithms are being developed for the following application fields:

• Model-based video content analysesfor security applications
• High performance content analyses for print inspection
• 3D vision for autonomous systems