Adaptive Segmentation Algorithm for Subtle Defect Images on the Surface of Magnetic Ring Using 2D-Gabor Filter Bank

Area of Science:

• Image processing and computer vision
• Materials science and engineering
• Artificial intelligence and machine learning

Background:

• Accurate segmentation of subtle defects on small magnetic rings is crucial for quality control.
• Existing unsupervised segmentation methods often struggle with noise and computational efficiency.

Purpose of the Study:

• To develop an adaptive threshold segmentation method for unsupervised segmentation of subtle defects on small magnetic rings.
• To improve segmentation accuracy and computational efficiency compared to traditional methods.

Main Methods:

• Utilized an improved multi-scale and multi-directional 2D-Gabor filter bank for noise reduction and feature enhancement.
• Analyzed grayscale statistical characteristics to construct an adaptive segmentation threshold.
• Employed a BP neural network classifier to differentiate between scar and crack defect types.

Main Results:

• The proposed adaptive threshold segmentation method achieved the highest classification accuracy at 97.5%.
• Demonstrated superior performance over iterative, OTSU, and maximum entropy methods in terms of accuracy and speed.
• Verified effective noise suppression and real-time processing capabilities on the SEED-DVS8168 platform.

Conclusions:

• The developed adaptive threshold segmentation method offers a robust and efficient solution for detecting subtle defects on small magnetic rings.
• The algorithm provides a significant advancement in image-based quality inspection for magnetic components.
• The method's high accuracy and computational efficiency make it suitable for real-time industrial applications.