Matching pursuit-based region-of-interest image coding.

Area of Science:

• Digital Signal Processing
• Image Compression
• Computer Vision

Background:

• Matching Pursuit (MP) is a signal analysis technique.
• MP can render image regions at specific qualities.
• Existing methods may lack scalability or adaptability.

Purpose of the Study:

• To present a novel, scalable, and progressive MP-based image coding scheme.
• To enable adaptive trade-offs between rate, distortion, and complexity.
• To allow interactive information refinement for prioritized image regions.

Main Methods:

• Utilizing Matching Pursuit (MP) for multiresolution signal analysis.
• Implementing a region-of-interest (ROI) coding scheme.
• Adapting MP analysis complexity by selecting subsets of the MP dictionary.

Main Results:

• The proposed scheme offers a controllable trade-off between rate, distortion, and complexity.
• It supports progressive information refinement for high-priority image regions.
• Computational complexity can be adapted to the encoder's capabilities.

Conclusions:

• The MP-based ROI image coding scheme is scalable and progressive.
• It provides interactive control over image quality and coding parameters.
• The method efficiently manages computational complexity for diverse hardware.