Robust and accurate sub-pixel extraction method of laser stripes in complex circumstances

Area of Science:

• 3D Laser Scanning Measurement
• Computer Vision
• Metrology

Background:

• Sub-pixel extraction of laser stripes is critical for 3D laser scanning accuracy.
• Challenges include noise separation and feature extraction in complex environments.
• Existing methods struggle with accuracy and robustness under difficult conditions.

Purpose of the Study:

• To develop a robust and accurate two-step method for sub-pixel feature extraction of laser stripes.
• To improve the precision and reliability of 3D laser scanning measurements.
• To address limitations of traditional methods in complex measurement scenarios.

Main Methods:

• Laser stripe segmentation using a deep semantic segmentation network for noise elimination.
• Sub-pixel extraction of gray peak points via Shepard sub-pixel interpolation and gray surface fitting.
• Utilizes gray distribution for high-precision and anti-interference results.

Main Results:

• The proposed method achieves more complete, denser, and smoother results compared to classical methods.
• Demonstrates superior performance in challenging conditions: large curved surfaces, high reflectivity, and intense ambient light.
• Experimental verification confirms robustness, effectiveness, and accuracy meeting high-precision measurement requirements.

Conclusions:

• The novel two-step approach significantly enhances sub-pixel laser stripe feature extraction.
• It provides a reliable solution for high-precision 3D laser scanning in complex environments.
• The method offers improved accuracy and anti-interference capabilities over traditional techniques.