Wyner-Ziv coding of video
- For numerous video compression scenarios, encoding is performed on low-power or mobile devices. Examples of such systems include wireless video sensors for surveillance, wireless PC cameras and mobile phones. For these applications, encoding must be implemented at the camera where memory and computation are scarce and low power usage is essential. Current video compression standards, such as MPEG or H.264/AVC, exploit the redundancy among video frames at the encoder. Due to the interframe processing performed at the encoder, these compression schemes are less suitable for applications which require low-complexity encoders. The Slepian-Wolf theorem on distributed source coding and the Wyner-Ziv results on source coding with decoder side information suggest that we can achieve efficient compression by exploiting the source statistics only at the decoder. Motivated by these theoretical results, a Wyner-Ziv video codec is developed. In this compression scheme individual frames are encoded independently, but decoded conditionally on adjacent reconstructed frames. Independently encoding the frames leads to low computation and memory requirements at the encoder, while the interframe processing at the decoder allows for efficient compression. As a building block for the Wyner-Ziv video codec, a practical scheme for compression with decoder side information based on turbo codes is presented. This coding technique performs close to the Slepian-Wolf bound and allows for rate-adaptivity and joint source-channel decoding. The Wyner-Ziv video codec is implemented in the pixel domain and the transform domain. Simulation results show that this video compression scheme achieves better rate-distortion performance than conventional low-complexity encoding schemes. To further improve the performance of the Wyner-Ziv video codec, two extensions are presented. Specifically, we implement hash-based motion estimation at the receiver. This allows the system to effectively estimate the motion at the decoder. We also extend the system to Wyner-Ziv coding of the residual images of a video sequence. With this scheme, the encoder uses a simple reference frame to calculate the residual while the decoder performs more sophisticated motion-compensation to generate side information for conditional decoding.
|Type of resource
|electronic; electronic resource; remote
|1 online resource.
|Aaron, Anne Margot Fernandez
|Stanford University, Department of Electrical Engineering
|Goldsmith, Andrea, 1964-
|Goldsmith, Andrea, 1964-
|Statement of responsibility
|Anne Margot F. Aaron.
|Submitted to the Department of Electrical Engineering.
|Ph. D. Stanford University 2011
- © 2011 by Anne Margot Fernandez Aaron
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...