| Bearbeiter | (nur für Mitarbeiter:innen einsehbar) |
| Betreuer | Dr. Jens Brandt |
| Professor | Prof. Dr.-Ing. Lars Wolf |
| IBR Gruppe | CM (Prof. Wolf) |
| Art | Masterarbeit |
| Status | abgeschlossen |
| Beginn | 2007-04-30 |
 Video adaptation is one of the major issues for video transmission in future wireless networks. Increasing bandwidth and processing power make it possible to receive video streams almost anywhere. However, the range of mobile devices varies from multimedia enabled mobile phones to high-end notebooks. We therefore have to deal with a diverse set of devices that vary in terms of their capabilities such as display resolution, processing power, memory size or network connection. A promising way to adapt to these circumstances for compressed digital video streams is the use of transcoding techniques. Transcoding techniques allow resource efficient video manipulation due to its application in the compressed domain. To support a wide set of video-enabled devices with different capabilities, video adaptation in more than one dimension is needed. Task DescriptionThe task of this thesis is to develop rate-control mechanisms for multidimensional transcoding. Firstly, rate-control mechanisms are required to provide video at a desired bandwidth. Secondly the transcoded video should also be of high quality. High quality is of subjective nature and can be described by several parameters, that are derived from the device's capabilities and the video itself. Thus, the structure and characteristics of the video should be taken into account when defining the transcoding parameters in each transcoding dimension. Therefore, methods for automatic identification of scene characteristics need to be investigated, in order to define transcoding parameters providing the highest quality. Current knowledge on quality definition should be observed. Besides these theoretical considerations practical limitations of the MPEG-4 standard need to be taken into account. Hence, different coding methods due to changing movie characteristics such as speed, complexity, possible cuts, zooms and in-scene movements have to be analysed. In a previous work a multidimensional transcoder for MPEG-4 video streams was developed. Based on this implementation, the developed rate control mechanisms have to be implemented as a controller module. It should analyse the video stream which is currently processed by the transcoder and determine based on some target device specifications some optimal settings to control the transcoding process automatically. Finally, the implementation has to be tested and evaluated. Links | |