An entropy-coded lattice vector quantizer for transform and subband image coding

Area of Science:

• Digital Signal Processing
• Image Compression
• Information Theory

Background:

• Traditional image coding methods often rely on complex vector codebooks.
• Efficient and high-performance image compression remains a significant challenge.

Purpose of the Study:

• To propose a novel lattice-based vector quantizer (VQ) and noiseless coding scheme.
• To develop a computationally efficient image coding method without requiring stored vector codebooks.

Main Methods:

• Implementation of a lattice-based vector quantizer for transform and subband image coding.
• Development of a noiseless code that enumerates lattice codevectors using their weighted L1 norm.
• Software implementation capable of handling large lattice codebooks (size 2^256).

Main Results:

• The proposed quantization method is simple to implement and eliminates the need for storing vector codebooks.
• The noiseless code effectively utilizes lattice codevector properties for efficient representation.
• Achieved image coding performance comparable or superior to state-of-the-art encoding methods.

Conclusions:

• The lattice-based VQ and noiseless coding offer a promising approach for advanced image compression.
• This method provides a practical and efficient alternative to existing image encoding techniques.
• The approach demonstrates high performance and implementation simplicity for transform and subband image coding.