Research on Defect Detection Method of Fusion Reactor Vacuum Chamber Based on Photometric Stereo Vision.

Area of Science:

• Robotics and Automation
• Computer Vision
• Nuclear Engineering

Background:

• Inspecting components within nuclear fusion reactor vacuum chambers presents challenges due to low-light conditions.
• Accurate 3D reconstruction is crucial for detecting defects on target plates.

Purpose of the Study:

• To develop and validate image enhancement and 3D reconstruction techniques for dim environments within nuclear fusion reactors.
• To enable precise defect detection and dimensional analysis of internal components.

Main Methods:

• An adaptive weighted multi-scale Retinex algorithm was employed for low-light image enhancement.
• A photometric stereo vision algorithm was utilized for 3D defect reconstruction.
• A light source illumination profile simulation system optimized light array placement.
• A robotic platform with a binocular stereo-vision camera was constructed for experiments.

Main Results:

• The image enhancement method successfully broadened the gray level, improving brightness and contrast.
• The 3D reconstruction achieved a maximum depth error below 24.0% and width error below 15.3%.
• The integrated system demonstrated effective defect detection and reconstruction capabilities.

Conclusions:

• The proposed techniques significantly improve visibility and dimensional accuracy for internal inspection in nuclear fusion reactors.
• This approach provides a viable solution for automated defect detection and 3D reconstruction in challenging, low-light environments.