An Efficient Residual-Based Method for Railway Image Dehazing.

Area of Science:

• Computer Vision
• Artificial Intelligence
• Transportation Safety

Background:

• Train operations are vulnerable to environmental factors like haze, which impairs visibility.
• Blurred vision due to haze poses significant risks to railway monitoring and operational safety.

Purpose of the Study:

• To develop an effective image dehazing method specifically for railway environments.
• To enhance the safety and reliability of train operations under hazy conditions.

Main Methods:

• Proposed an end-to-end residual block-based network (RID-Net) with fine-grained and coarse-grained subnetworks.
• Utilized a combined loss function incorporating per-pixel and perceptual losses for high-quality image restoration.
• Evaluated performance using PSNR, SSIM, object detection, running time, and sensory vision.

Main Results:

• RID-Net demonstrated superior performance in image dehazing compared to state-of-the-art methods.
• The method achieved high-quality restored images by preserving both low-level and high-level features.
• Effective dehazing was confirmed across synthesized railway, indoor, and real-world datasets.

Conclusions:

• RID-Net offers a robust solution for railway image dehazing, enhancing operational safety.
• The proposed method significantly improves visibility in hazy conditions, crucial for railway monitoring.
• This advancement contributes to safer and more efficient railway transportation systems.