The Detection of Yarn Roll's Margin in Complex Background.

Area of Science:

• Textile Engineering
• Computer Vision
• Automation Systems

Background:

• Online detection of yarn roll margins is crucial for efficient textile automation, impacting bobbin replacement scheduling.
• Industrial environments present challenges like uneven lighting, varied yarn colors, complex backgrounds, and restricted shooting angles, leading to detection errors.
• Existing neural network and contour detection methods struggle with accuracy due to these environmental complexities.

Purpose of the Study:

• To develop an improved algorithm for accurate online detection of yarn roll margins in textile automation.
• To enhance the precision of yarn roll dimension measurement (diameter and length) under adverse industrial conditions.
• To integrate Yolo and contour detection with Kalman filtering for robust and error-free yarn volume estimation.

Main Methods:

• An improved Yolo algorithm was integrated with a contour detection algorithm for initial yarn roll and dimension detection.
• Fusion of diameter measurements from Yolo and contour detection, followed by calculation of yarn roll length and edges.
• Application of a Kalman filter to fuse measured and estimated residual yarn volume (based on yarn consumption speed) for error elimination.

Main Results:

• The integrated algorithm successfully detects yarn rolls and their dimensions in complex industrial settings.
• The fused measurement approach significantly improves the accuracy of yarn roll diameter and length calculations.
• Experimental verification shows an average measurement error of less than 8.6 mm for cylinder yarn diameter and under 3 cm for cylinder yarn length.

Conclusions:

• The proposed improved Yolo and contour detection algorithm effectively overcomes challenges of complex backgrounds and illumination in textile automation.
• This method provides a robust solution for accurate online yarn roll margin detection and dimension measurement.
• The integration with Kalman filtering ensures reliable estimation of residual yarn volume, applicable to diverse yarn roll types.