Postprocessing of low bit-rate block DCT coded images based on a fields of experts prior

Area of Science:

• Digital image and video processing
• Signal processing and communications

Background:

• Transform coding, particularly Discrete Cosine Transform (DCT), is prevalent in image and video standards.
• Low bit rates in transform coding lead to significant visual distortions, limiting compression efficiency.
• Postprocessing techniques are crucial for mitigating these distortions and balancing bit rate reduction with quality preservation.

Purpose of the Study:

• To develop an advanced postprocessing method for transform-coded images to reduce visual artifacts.
• To address the trade-off between bit rate reduction and image quality preservation in image and video coding.

Main Methods:

• The proposed method treats postprocessing as an inverse problem, solved using the maximum a posteriori (MAP) criterion.
• Image distortion from coding is modeled as additive, spatially correlated Gaussian noise.
• The original image is modeled using a high-order Markov random field (MRF) within the fields of experts (FoE) framework.

Main Results:

• The MAP-based postprocessing method significantly reduces visual distortions in low bit rate coded images.
• Experimental results demonstrate superior performance compared to existing methods, achieving higher Peak Signal-to-Noise Ratio (PSNR) gains.
• The processed images exhibit improved visual quality.

Conclusions:

• The proposed MAP-based approach effectively enhances the visual quality of transform-coded images at low bit rates.
• The study validates the effectiveness of the chosen noise and image models for postprocessing applications.
• The research provides insights into noise model assumptions and parameter settings, addressing potential issues with coefficient truncation.