Polarization motivating high-performance weak targets' imaging based on a dual-discriminator GAN

Area of Science:

• Computer Vision
• Image Processing
• Artificial Intelligence

Background:

• Detecting weak targets under bright light is a significant challenge in various applications.
• Existing methods often struggle to balance target visibility with background preservation.

Purpose of the Study:

• To develop an effective image fusion method for enhancing weak target detection in bright light conditions.
• To leverage both intensity and polarization information for improved detection accuracy.

Main Methods:

• An attention-guided dual-discriminator generative adversarial network (GAN) was designed for fusing intensity and polarization images.
• The framework incorporates an attention mechanism to focus on contextual semantic information and enhance long-term dependencies.
• A specialized loss function was employed to preserve salient information and pixel-level distribution.

Main Results:

• The proposed fusion method successfully integrated rich background details from intensity images with crucial target information from polarization images.
• Experimental results on a real-scene dataset demonstrated superior performance compared to existing methods.
• The method showed effectiveness in detecting various weak targets under challenging bright light conditions.

Conclusions:

• The attention-guided GAN-based fusion method significantly improves the detection of weak targets in bright light.
• Combining intensity and polarization information through this advanced technique offers a robust solution for challenging visual tasks.
• The developed framework provides a promising direction for high-level target detection in adverse lighting.