Surface Defect Detection with Gabor Filter Using Reconstruction-Based Blurring U-Net-ViT

2025-08-31

AI Summary

To address the challenges of strong background interference and weak defect features in texture surface defect detection, this paper proposes a Gabor-enhanced hybrid U-Net–ViT model. It leverages Gabor filtering to pre-extract orientation- and frequency-sensitive texture priors, synergistically integrating U-Net’s local detail reconstruction capability with ViT’s global contextual modeling. A Gaussian-weighted Dice loss is introduced to improve segmentation boundary accuracy, and a salt-and-pepper masking strategy is designed to enhance edge perception of defects. Adaptive optimization of Gabor parameters and multi-scale feature fusion further boost robustness in localizing and segmenting minute defects. Evaluated on MVTec-AD, Surface Crack Detection, and Marble Surface Anomaly Dataset, the model achieves an average AUC of 0.939. Ablation studies confirm statistically significant contributions of each component to overall performance.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Adversarial Learning & Robustness
• Natural Language Processing: Safety and Robustness

Application Category

• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics

Abstract

This paper proposes a novel approach to enhance the accuracy and reliability of texture-based surface defect detection using Gabor filters and a blurring U-Net-ViT model. By combining the local feature training of U-Net with the global processing of the Vision Transformer(ViT), the model effectively detects defects across various textures. A Gaussian filter-based loss function removes background noise and highlights defect patterns, while Salt-and-Pepper(SP) masking in the training process reinforces texture-defect boundaries, ensuring robust performance in noisy environments. Gabor filters are applied in post-processing to emphasize defect orientation and frequency characteristics. Parameter optimization, including filter size, sigma, wavelength, gamma, and orientation, maximizes performance across datasets like MVTec-AD, Surface Crack Detection, and Marble Surface Anomaly Dataset, achieving an average Area Under the Curve(AUC) of 0.939. The ablation studies validate that the optimal filter size and noise probability significantly enhance defect detection performance.

Problem

• Enhancing surface defect detection accuracy using Gabor filters and blurring U-Net-ViT model
• Detecting defects across various textures by combining local and global feature processing
• Removing background noise and highlighting defect patterns in noisy industrial environments

Innovation

• Combining U-Net and Vision Transformer for defect detection
• Using Gaussian filter loss to remove background noise
• Applying Gabor filters to emphasize defect characteristics