RA2: Visual Computing

Vision and Research Strategy

Visual Computing deals with computer-aided representation and processing of visual information. It includes, in particular, the areas of image analysis and synthesis. While textual information processing by computers is a well-established and successful area, the quality, speed and robustness of many current visual computing algorithms is still far behind human capabilities and requires much more research. The goal of RA2 is to establish a coherent and robust bottom-up framework for a number of key problems in visual computing. It will be based on transparent first principles such as expressing computer vision problems in terms of optimization formulations. We will develop highly efficient real-time algorithms that are well-suited for new hardware platforms, and we will explore feature-based compact representations for visual data. Video-based rendering will play an important role as a testbed for many of our algorithms, and will serve as a demonstrator for a pipeline that unifies the previously separate disciplines of image analysis and image synthesis. Finally, we will explore ways to integrate the sensory information of text, speech and vision. We envision that this integration will enable the development of novel systems that have a robustness and flexibility that comes significantly closer to humans' performance.

Research Topics and Achievements

Responsible Investigators and Group Development


Prizes and Awards