A Multi-Scale Edge-Preserving Decomposition and Fusion Framework for Multi-Polarization Passive Millimeter-Wave...

Area of Science:

• Imaging Technology
• Signal Processing
• Security Screening

Background:

• Passive millimeter-wave (PMMW) imaging offers privacy-preserving human body security inspections.
• Existing PMMW methods struggle with false alarms and missed detections due to single-pixel/polarization processing.
• Current multi-polarization techniques inadequately exploit rich distinguishing features.

Purpose of the Study:

• To develop a systematic framework for fully exploiting multi-polarization information in PMMW imaging.
• To enhance detection accuracy and robustness in PMMW-based security inspections.
• To improve the contrast of concealed objects in PMMW images.

Main Methods:

• Proposed a multi-scale edge-preserving decomposition model: Gaussian and weighted average curvature filtering (GWACF).
• Hierarchically decomposed PMMW images into base structural (BS), coarse structural (CS), and fine structural (FS) layers.
• Developed a fusion strategy using gradient-domain pulse-coupled neural network (PCNN) and energy attribute fusion.

Main Results:

• The GWACF model effectively decomposed multi-polarization PMMW images.
• The proposed fusion strategy leveraged complementary polarimetric information without artifacts or edge degradation.
• Significantly enhanced brightness temperature (BT) contrast of concealed objects.

Conclusions:

• The novel PMMW imaging method demonstrates superior performance compared to existing techniques.
• The approach offers notable advantages in detection accuracy and robustness for security applications.
• This work provides a systematic framework for advanced multi-polarization PMMW image analysis.