Small Defects Detection of Galvanized Strip Steel via Schatten-p Norm-Based Low-Rank Tensor Decomposition

TL;DR

A new method for detecting small defects on galvanized steel is simpler and more effective than deep learning, with potential use in other industries.

Contribution

A Schatten-p norm-based low-rank tensor decomposition method for small defect detection in industrial settings.

Findings

The proposed SLRTD method outperforms existing methods in detecting white-spot defects on galvanized steel.

The method effectively separates defect images into low-rank background, sparse defect, and noise components.

The approach is applicable to other industrial products like glass, fabric, and LCDs.

Abstract

What are the main findings?
Compared with the deep learning method for surface defect detection, the proposed SLRTD-based detection method is simpler and more effective in a galvanized strip steel production line.

Compared with the deep learning method for surface defect detection, the proposed SLRTD-based detection method is simpler and more effective in a galvanized strip steel production line.

What is the implication of the main finding?
The proposed SLRTD-based detection method of surface defect can be applied for other industrial products, such as glass, fabric, LCD, and AMOLED.

The proposed SLRTD-based detection method of surface defect can be applied for other industrial products, such as glass, fabric, LCD, and AMOLED.

Accurate and efficient white-spot defects detection for the surface of galvanized strip steel is one of the most important guarantees for the quality of steel production. It is a fundamental but “hard” small target detection problem due to its small pixel occupation in low-contrast images. By fully exploiting the low-rank and sparse prior information of a surface defect image, a Schatten-p norm-based low-rank tensor decomposition (SLRTD) method is proposed to decompose the defect image into low-rank background, sparse defect, and random noise. Firstly, the original defect images are transformed into a new patch-based tensor mode through data reconstruction for mining valuable information of the defect image. Then, considering the over-shrinkage problem in the low-rank component estimation caused by a vanilla nuclear norm and a weighted nuclear norm, a nonlinear reweighting strategy based on a Schatten p-norm is incorporated to improve the decomposition performance. Finally, a solution framework is proposed via a well-designed alternating direction method of multipliers to obtain the white-spot defect target image by a simple segmenting algorithm. The white-spot defect dataset from a real-world galvanized strip steel production line is constructed, and the experimental results demonstrate that the proposed SLRTD method outperforms existing state-of-the-art methods qualitatively and quantitatively.

Figures