Related Experiment Videos
Bayesian resolution enhancement of compressed video
C Andrew Segall1, Aggelos K Katsaggelos, Rafael Molina
1Department of Electrical and Computer Engineering, Northwestern University, Evanston, IL 60208, USA. asegall@ece.nwu.edu
Summary
This study integrates video compression effects into super-resolution algorithms. We establish relationships between compression, resolution recovery, and algorithm parameters for enhanced performance.
Area of Science:
- Digital Image Processing
- Video Compression
- Super-Resolution Algorithms
Background:
- Super-resolution algorithms enhance image detail from low-resolution inputs.
- Video compression techniques impact the quality and information available for super-resolution.
Purpose of the Study:
- To investigate the effects of video compression on super-resolution performance.
- To develop a unified framework for super-resolution and post-processing under compression.
Main Methods:
- Utilizing a Bayesian framework to incorporate information from hybrid motion-compensation and transform coding.
- Defining a tractable solution that fuses super-resolution and post-processing.
- Establishing relationships between algorithm parameters and compressed bitstream information.
Main Results:
- Demonstrated a method to fuse super-resolution and post-processing problems within a Bayesian framework.
- Established clear associations between resolution recovery and compression ratio.
- Identified relationships between algorithm parameters and compressed bitstream data.
Conclusions:
- The proposed method effectively integrates video compression information into super-resolution.
- Performance is validated through simulations on synthetic and real-world sequences.
- Understanding compression's impact is crucial for optimal super-resolution.