Three-dimensional photometric reconstruction based automated air-void segmentation system for hardened concrete

Description:

The system comprises a hardware setup for capturing high-resolution images of concrete specimens and a software sub-system for data processing and 3D reconstruction. Utilizing a Sony α 7R II camera and six LED lights, the hardware captures images at a resolution capable of identifying air voids as small as 10 μm. The software employs photometric stereo methods and a U-Net based AI model for accurate 3D reconstruction and air-void segmentation of concrete surfaces.

Key Advantages:

• High-resolution imaging capable of capturing minute air voids.
• Automated and accurate air-void segmentation using AI.
• Effective in various lighting conditions due to controlled LED lighting.
• Minimized ambient light disturbance for consistent image quality.
• Capable of handling different concrete specimens for comprehensive analysis.

Problems Solved:

• Challenges in detecting small air voids in concrete specimens.
• Manual and time-consuming methods of air-void segmentation.
• Inaccuracies in 3D reconstruction of concrete surfaces.
• Difficulty in distinguishing air voids from similar appearances in images.

Market Applications:

• Quality control and analysis in concrete manufacturing.
• Research and development in construction materials.
• Structural health monitoring for infrastructure.
• Enhancement of automated image segmentation in construction technology.