Merge Frame Design for Interactive Media Streaming Applications
Event details
| Date | 03.03.2016 |
| Hour | 11:00 › 12:30 |
| Speaker | Prof. Gene Cheung, NII, Tokyo |
| Location | |
| Category | Conferences - Seminars |
The advent of multi-camera / light field capturing technologies means that the acquired image / video data can be in much higher dimension than a user is able to observe on his conventional display. This means a user must navigate through the high dimensional data one image at a time, resulting in an interactive streaming paradigm: a user periodically requests one data subset from server, and the server must transmit pre-encoded data for correct decoding of the requested media. Achieving good compression efficiency for interactive streaming is difficult, because the image navigation path by a user is not known at encoding time, making differential coding difficult to employ.
We propose a new distributed source coding (DSC) approach to support high-dimensional image navigation. Specifically, we design a merge frame (M-frame) to “merge” multiple side information (SI) frames--each can possibly be available at the decoder buffer due to different navigation paths—into a unique frame reconstruction to facilitate differential coding. We use a piecewise constant (PWC) function as the merge operator, and transmit a series of parameters of these PWC merge functions in order to guarantee identical reconstruction given the known SI blocks. We consider two different scenarios.
In the first case, a target frame is first given, and then merge parameters are chosen so that this frame can be reconstructed exactly at the decoder. In the second scenario, the reconstructed frame and merge parameters are jointly optimized to meet a rate-distortion criteria. Experiments show that for both scenarios, our proposed merge techniques can outperform both a recent DSC approach using bit-plane and channel coding and the SP-frame approach in H.264, in terms of compression efficiency and decoder complexity. Finally, we discuss how the proposed M-frame can be used to design frame structures for an interactive light field streaming application.
We propose a new distributed source coding (DSC) approach to support high-dimensional image navigation. Specifically, we design a merge frame (M-frame) to “merge” multiple side information (SI) frames--each can possibly be available at the decoder buffer due to different navigation paths—into a unique frame reconstruction to facilitate differential coding. We use a piecewise constant (PWC) function as the merge operator, and transmit a series of parameters of these PWC merge functions in order to guarantee identical reconstruction given the known SI blocks. We consider two different scenarios.
In the first case, a target frame is first given, and then merge parameters are chosen so that this frame can be reconstructed exactly at the decoder. In the second scenario, the reconstructed frame and merge parameters are jointly optimized to meet a rate-distortion criteria. Experiments show that for both scenarios, our proposed merge techniques can outperform both a recent DSC approach using bit-plane and channel coding and the SP-frame approach in H.264, in terms of compression efficiency and decoder complexity. Finally, we discuss how the proposed M-frame can be used to design frame structures for an interactive light field streaming application.
Links
Practical information
- Informed public
- Free
- This event is internal
Organizer
- LTS4 / APSIPA distinguished lecture series
Contact
- Prof. Pascal Frossard