Abstract
This paper introduces two polynomial regression solutions for error concealment by predicting the values of motion vectors (MVs) of lost macroblocks. The two solutions are online and offline polynomial regression modeling. In the former solution, the regression model is built during the decoding process, while in the latter solution, the model is built during the encoding or the transcoding process and then used at the decoder for concealment. Both solutions make use of the spatially and temporally neighboring MVs for building the regression models. The advantages and disadvantages of the proposed solutions are elaborated upon. In comparison with existing work, the experimental results show that the proposed solutions have clear advantages of computational time requirements and MV prediction accuracy.
| Original language | English |
|---|---|
| Pages (from-to) | 581-588 |
| Number of pages | 8 |
| Journal | Signal, Image and Video Processing |
| Volume | 9 |
| Issue number | 3 |
| DOIs | |
| State | Published - Mar 2015 |
| Externally published | Yes |
Keywords
- Error concealment
- Machine learning
- Regression
- Video compression
Fingerprint
Dive into the research topics of 'H.264/AVC motion vector concealment solutions using online and offline polynomial regression'. Together they form a unique fingerprint.Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver