Magnesium Ingot Stacking Segmentation Algorithm for Industrial Robot Based on the Correction of Image Overexposure...

Area of Science:

• Materials Science
• Computer Vision
• Image Processing

Background:

• Magnesium alloy ingot surface inspection is crucial for quality control.
• Mirror reflections from lighting obscure critical surface details.
• Existing segmentation methods struggle with overexposed regions.

Purpose of the Study:

• To develop an adaptive threshold segmentation algorithm for magnesium ingots.
• To address and correct overexposed areas caused by reflections.
• To enhance the accuracy of magnesium alloy ingot segmentation.

Main Methods:

• Adaptive Threshold Image Overexposure area Correction (ATSIOAC) algorithm.
• Dividing exposure probability into weak and strong regions based on brightness and chromaticity.
• Using saturation difference for magnesium ingot region masking.
• Correcting overexposed areas using RGB average of adjacent pixels.
• Applying pixel weighted average (WA) for smooth image fusion.

Main Results:

• Effective correction of color information in overexposed areas.
• Complete segmentation of the top surface of magnesium ingot piles.
• Significant improvement in magnesium alloy ingot segmentation accuracy.
• Achieved a segmentation accuracy of 94.38%.

Conclusions:

• The proposed ATSIOAC algorithm successfully corrects overexposed regions in magnesium ingot images.
• The algorithm enhances segmentation accuracy for quality inspection.
• This method provides a robust solution for automated inspection of magnesium alloy ingots.