Ultra-High Voltage GIS Equipment X-ray Image Enhancement and Stitching

This paper presents a comprehensive image processing framework for the enhancement and stitching of X-ray images of ultra-high voltage GIS equipment. To address geometric misalignment caused by minor shifts and rotations during imaging, a robust affine transformation model estimated via RANSAC is introduced for accurate image registration. Additionally, a brightness correction strategy based on matched feature points and a linear blending method are employed to ensure seamless transitions in overlapping regions. Experimental results demonstrate that the proposed method significantly improves both structural consistency and visual quality. The study offers a practical and effective solution for industrial non-destructive testing, with potential applications in automated defect detection and system health monitoring.