Abstract
In this paper, we specifically design an efficient compressive sensing video (CSV) coding framework for the CSV system, by considering the distribution characteristics of the CSV frame. To explore the spatial redundancy of the CSV, the CSV frame is first divided into blocks and each block is modeled by a Gaussian mixture model (GMM), and then it is compressed by a product vector quantization. We further explore the temporal redundancy of the CSV by encoding the adjacent CSV frames by the differential pulse code modulation technique and the arithmetic encoding technique. Experiment results show that the proposed CSV coding solution maintains low coding complexity, which is required by the CSV system. Meanwhile, it achieves significant BD-PSNR improvement by about 7.13–11.41 dB (or equivalently 51.23–66.96% bitrate savings) compared with four existing video coding solutions, which also have low computational complexity and suit for the CSV system.
| Original language | English |
|---|---|
| Pages (from-to) | 66-79 |
| Number of pages | 14 |
| Journal | Signal Processing: Image Communication |
| Volume | 55 |
| DOIs | |
| State | Published - 1 Jul 2017 |
Keywords
- Compressive sensing video
- Gaussian mixture model
- Lossy compression
- Product vector quantizer
- Video coding
Fingerprint
Dive into the research topics of 'A new compressive sensing video coding framework based on Gaussian mixture model'. Together they form a unique fingerprint.Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver