A texture guided transmission line image enhancement method
View abstract on PubMed
Summary
This summary is machine-generated.A new texture-guided transmission line image enhancement (TGTLIE) method improves foreign object detection accuracy. This AI-powered approach effectively enhances image quality degraded by rain, fog, and blur for better transmission line inspection.
Area Of Science
- Computer Vision
- Artificial Intelligence
- Image Processing
Background
- Transmission line inspection relies heavily on image detection for foreign object identification.
- Environmental factors like rain, fog, and blur degrade image quality, hindering detection accuracy.
Purpose Of The Study
- To propose a novel image enhancement method for transmission line inspection.
- To improve the accuracy of foreign object detection on transmission lines despite environmental interference.
Main Methods
- A texture inference network (TINet) extracts texture information.
- A texture-based conditional generative adversarial network (TCGAN) performs adaptive deraining, defogging, and deblurring.
- A neural gradient algorithm, dual path attention, and a global-local discriminator enhance image generation and prevent artifacts.
Main Results
- The TGTLIE method effectively removes noise and enhances image quality under various conditions.
- Achieved high PSNR (up to 34.921 dB) and SSIM (up to 0.962) values.
- Demonstrated excellent performance in foreign object detection tasks.
Conclusions
- The proposed TGTLIE method offers robust image enhancement for transmission line inspection.
- Provides effective technical support for intelligent inspection and fault warning systems.
- Significantly improves the reliability of automated visual inspection in challenging environments.