Removal of compression artifacts using projections onto convex sets and line process modeling

Area of Science:

• Digital image processing
• Computer vision
• Signal processing

Background:

• Compression algorithms like MPEG introduce blocking and ringing artifacts.
• Existing artifact removal methods may not effectively handle spatially varying artifact visibility.

Purpose of the Study:

• To develop a new image recovery algorithm for enhanced artifact removal.
• To address both blocking and ringing artifacts in compressed images and video.

Main Methods:

• Utilized the theory of projections onto convex sets (POCS).
• Introduced directional smoothness constraint sets based on line processes for edge modeling.
• Incorporated a divide-and-conquer (DAC) strategy to simplify projection computations.
• Accounted for spatially varying artifact visibility in the smoothness set definitions.

Main Results:

• Demonstrated effective removal of blocking and ringing artifacts.
• Numerical experiments confirmed the algorithm's performance with MPEG-based codecs.
• The DAC strategy improved the computational efficiency of projections.

Conclusions:

• The proposed POCS-based algorithm successfully removes blocking and ringing artifacts.
• The novel directional smoothness constraints and DAC strategy offer an effective approach to image recovery.
• The algorithm shows promise for improving the quality of compressed visual media.