Hybrid order statistic filter and its application to image restoration

Area of Science:

• Signal Processing
• Image Restoration
• Nonlinear Filtering

Background:

• Traditional linear filters struggle with preserving edges and suppressing certain noise types.
• Existing nonlinear filters offer advantages but may not handle all noise scenarios effectively.

Purpose of the Study:

• To introduce and evaluate a novel nonlinear filter, the hybrid order statistic (HOS) filter.
• To demonstrate the HOS filter's capability in signal and image restoration tasks.

Main Methods:

• Development of the hybrid order statistic (HOS) filter, integrating rank- and spatial-order information.
• Comparative analysis against linear Wiener and nonlinear L, Ll, and rank-conditioned rank selection filters.

Main Results:

• The HOS filter demonstrates superior performance in reducing mean-squared error for Gaussian noise-contaminated signals.
• It effectively preserves edges and reduces impulsive noise while suppressing Gaussian noise.

Conclusions:

• The HOS filter offers a robust solution for signal and image restoration.
• It combines the benefits of linear and nonlinear filtering approaches for enhanced performance.