Underwater Structured Light Stripe Center Extraction with Normalized Grayscale Gravity Method

Area of Science:

• Robotics and Automation
• Computer Vision
• Optical Engineering

Background:

• Underwater laser measurements face challenges due to non-uniform surface reflectance and environmental disturbances.
• Accurate laser stripe center extraction is crucial for 3D reconstruction and analysis in underwater environments.

Purpose of the Study:

• To develop a robust method for precise underwater laser stripe center extraction.
• To address the impact of non-uniform reflectance and environmental disturbances on measurement accuracy.

Main Methods:

• Building a comprehensive underwater structured light dataset with varying conditions.
• Comparing state-of-the-art semantic segmentation models, recommending PSPnet.
• Implementing a normalized grayscale gravity method with enhanced robustness and subpixel accuracy via bilinear interpolation.

Main Results:

• The proposed normalized grayscale gravity method effectively eliminates noise interference.
• The method demonstrates good robustness and self-adaptability in diverse underwater conditions.
• PSPnet is identified as a suitable model for underwater structured light stripe segmentation.

Conclusions:

• The developed method significantly improves laser stripe center extraction accuracy in underwater settings.
• The approach offers a reliable solution for challenging underwater optical measurements.
• The study contributes to advancing underwater 3D sensing technologies.