JPEG-compliant perceptual coding for a grayscale image printing pipeline.

Area of Science:

• Digital Image Processing
• Computer Vision
• Information Theory

Background:

• Joint Photographic Experts Group (JPEG) compression is widely used for image compression.
• Standard JPEG compression may introduce artifacts, especially when images are scaled, halftoned, and printed.
• Optimizing compression for specific image processing pipelines can improve visual quality and efficiency.

Purpose of the Study:

• To develop a customized JPEG compression procedure for grayscale images.
• To maintain compatibility with the existing JPEG standard while enhancing performance.
• To reduce compression artifacts and improve visual quality in scaled, halftoned, and printed images.

Main Methods:

• Designed JPEG quantization tables using frequency-domain characteristics of scaling and halftoning.
• Incorporated human visual system's frequency sensitivity into table design.
• Optimized Huffman tables for low-rate coding.
• Developed an iterative procedure to account for scaling and halftoning effects.

Main Results:

• Achieved significant reduction in compression artifacts by masking them with halftoning patterns.
• Pushed artifacts into frequency bands less sensitive to human vision.
• Maintained near visually lossless image quality at rates below 0.5 bits per pixel.
• Demonstrated at least 20% better performance compared to standard JPEG encoders.

Conclusions:

• The customized JPEG compression method is compatible with all scaling and halftoning techniques.
• The approach significantly enhances image quality and compression efficiency for print workflows.
• This method offers a substantial improvement over standard JPEG compression for specific applications.