Lapped multiple bases algorithms for still image compression without blocking effect

Area of Science:

• Digital Signal Processing
• Image Processing
• Computer Vision

Background:

• Traditional image compression methods, such as Discrete Cosine Transform (DCT), face limitations in achieving high compression ratios without introducing visual artifacts like blocking effects.
• The need for efficient image compression is critical for applications with limited bandwidth or storage, driving research into novel transform-based algorithms.

Purpose of the Study:

• To develop and evaluate a novel still image compression system utilizing multiple transform sets within a multiple bases realization algorithm.
• To enhance compression efficiency beyond existing DCT-only methods.
• To explore the efficacy of newly developed lapped orthogonal transform (LOT) bases for low bit rate image compression.

Main Methods:

• Implementation of a multiple bases realization algorithm incorporating several transform sets.
• Comparison of the proposed algorithm's performance against DCT-only compression in terms of encoded transform coefficients.
• Integration and assessment of newly developed lapped orthogonal transform (LOT) bases within the compression framework.

Main Results:

• The developed algorithms achieved a 20% reduction in encoded transform coefficients compared to DCT-only compression.
• The extended algorithms using LOT bases demonstrated effective performance at low bit rates (high compression).
• The use of LOT bases successfully mitigated the blocking effect, a common artifact in high compression scenarios.

Conclusions:

• The multiple bases realization algorithm offers a significant improvement in still image compression efficiency.
• The incorporation of LOT bases provides a viable solution for high compression operation without introducing blocking artifacts.
• This research contributes advanced techniques for efficient and artifact-free image compression, particularly beneficial for low bit rate applications.