Image-Based Bolt-Loosening Detection Using a Checkerboard Perspective Correction Method

Area of Science:

• Engineering
• Computer Vision
• Mechanical Systems

Background:

• Flange joints are critical in mechanical systems, and bolt integrity is essential for structural safety.
• Accurate bolt status monitoring is challenging due to image distortions and camera positioning errors.

Purpose of the Study:

• To develop and validate a novel image-correction method for flange joint bolts.
• To enhance the stability and accuracy of bolt-loosening detection.
• To analyze the influence of various factors on the correction method's performance.

Main Methods:

• A checkerboard pattern is used to identify feature points on flange joint bolts.
• Homography matrix estimation using multiple feature points corrects image perspective distortion and camera angle deviation.
• Analysis of factors including feature point count, camera distance, and light intensity.

Main Results:

• The proposed method effectively corrects image distortion and camera position angle deviation.
• Using more than four correction points significantly improves the stability of bolt image correction and bolt-loosening detection.
• The method remains effective with as few as nine checkerboard points, achieving an average bolt-loosening detection error below 1.5 degrees.

Conclusions:

• The developed image-correction technique offers a robust solution for flange joint bolt analysis.
• The method demonstrates improved stability and accuracy in detecting bolt loosening.
• Optimal camera shooting distances range from 20 cm to 60 cm, with low sensitivity to light intensity.