Coherent Noise Suppression Using Adaptive Homomorphic Filtering for Wideband Electromagnetic Imaging System

Area of Science:

• Electromagnetic imaging
• Signal processing
• Interference detection

Background:

• Wideband electromagnetic imaging systems detect electromagnetic interference sources (EMIS).
• Coherent noise from multiple EMIS degrades image quality, posing challenges for conventional iterative denoising methods due to frequency-dependent noise patterns.
• Existing techniques struggle with wideband noise removal where patterns change with frequency.

Purpose of the Study:

• To propose an adaptive homomorphic filtering technique for effective coherent noise removal in wideband electromagnetic images (1 GHz-6 GHz).
• To address the limitations of conventional methods in handling frequency-varying coherent noise patterns.

Main Methods:

• Investigated coherent noise characteristics in electromagnetic images, linking pattern variations to EMIS position, number, and frequency.
• Developed an adaptive Gaussian filter based on coherent noise intensity probability density analysis.
• Utilized the minimum description length (MDL) criterion for adaptive filter parameter selection, enabling localized smoothing.

Main Results:

• The proposed adaptive homomorphic filtering method effectively removes coherent noise across a wide frequency band.
• Demonstrated significant improvements in electromagnetic image quality: a 15 dB reduction in maximum sidelobe level (MSL) and over 20 dB enhancement in dynamic range (DR).
• Outperformed conventional narrowband denoising methods in both experimental and simulation results.

Conclusions:

• Adaptive homomorphic filtering provides a robust solution for coherent noise suppression in wideband electromagnetic imaging.
• The developed method offers substantial improvements in image quality and system dynamic range, crucial for accurate EMIS detection and localization.