Improved MPEG-4 still texture image coding under noisy environment.

Area of Science:

• Digital Image Processing
• Video Compression Standards
• Signal Processing

Background:

• MPEG-4 still texture image codec performance with noisy images is not well understood.
• Compression artifacts, particularly the
• cross
• shape, degrade visual quality in noisy image compression.
• Existing methods may not effectively balance compression rate and image fidelity for noisy content.

Purpose of the Study:

• Analyze the causes of optimal operating points and
• cross
• artifacts in MPEG-4 compression of noisy images.
• Propose an adaptive thresholding technique to mitigate these artifacts.
• Evaluate the effectiveness of the proposed technique in improving visual quality and peak-signal-to-noise ratio (PSNR).

Main Methods:

• Analysis of MPEG-4 still texture image codec behavior with noisy input data.
• Development of an adaptive thresholding algorithm integrated into the MPEG-4 quantization process.
• Analytical study for threshold value selection and validation.
• Comparative evaluation of the proposed method against the traditional MPEG-4 codec.

Main Results:

• Identified an optimal operating point for peak-signal-to-noise ratio (PSNR) in low bit rate regions for noisy images.
• Characterized the
• cross
• shape artifact as a significant visual degradation issue.
• Demonstrated substantial improvements in visual quality and PSNR using the proposed adaptive thresholding technique.
• Confirmed that the decoder remains unchanged, simplifying implementation.

Conclusions:

• The proposed adaptive thresholding technique effectively removes
• cross
• artifacts and enhances image quality for noisy images compressed with MPEG-4.
• The method offers a practical solution with minimal modification to the encoder.
• This approach significantly improves upon the performance of the traditional MPEG-4 still texture image codec for noisy image coding.