Dr Patrick Ndjiki-Nya from the Fraunhofer Heinrich Hertz Institute, Germany, will be presenting the below seminar today. Please feel free to attend.
The seminar will be held at 11.10am in the basement, Ingkarni Wardli Building, B21, Teaching Suite 5.
Bio: Dr. Patrick Ndjiki-Nya received the Dipl.-Ing. title (corr. to M.S degree) from the Technical University of Berlin in 1997. In 2008 he also finished his doctorate at the Technische Universität Berlin. He has developed an efficient method for content-based video coding, which combines signal theory with computer graphics and vision. His approaches are currently being evaluated in equal or similar form by various companies and research institutions in Europe and beyond. From 1997 to 1998 he was significantly involved in the development of a flight simulation software at Daimler-Benz AG. From 1998 to 2001 he was employed as development engineer at DSPecialists GmbH where he was concerned with the implementation of algorithms for digital signal processors (DSP). During the same period he researched content-based image and video features at the Fraunhofer Institute for Telecommunications – Heinrich Hertz Institute, Berlin, Germany, with the purpose of implementation in DSP solutions from DSPescialists GmbH. Since 2001 he is solely employed at Fraunhofer Heinrich Hertz Institute, where he was Project Manager initially and Senior Project Manager from 2004 on. He has been appointed Head of Group in 2010. He currently coordinates several national and European projects related to 2D and 3D (semantic) content assessment.
Summary of Presentation: All major video coding standards since H.261 (1980’s) have been based on incremental improvements to the hybrid motion-compensated block transform coding model. While this approach has produced impressive rate-distortion improvements, it is believed that more disruptive techniques can provide substantial gains. The current H.264/AVC compression standard is based on the picture-wise processing and waveform-based coding of video signals.
The approach is underpinned by the assumption that, in most cases, the target of video compression is to provide good subjective quality rather than to minimise the error between the original and coded pictures. It is thus possible to conceive of a compression scheme where an texture analysis/synthesis framework replaces the conventional energy minimisation approach. Such a scheme could offer substantially lower bitrates through reduced residual and motion vector coding. New and alternative frameworks of this type are beginning to emerge, where prediction and signal representation are based on a parametric or data-driven model of scene content. These often invoke a combination of waveform coding and texture replacement, where computer graphic models are employed to replace target textures at the decoder. Work to date has demonstrated the potential for dramatic rate-quality improvements with such methods.
Please make time to attend the seminar if you can.