Hierarchical low-rank and sparse tensor micro defects decomposition by electromagnetic thermography imaging system

Area of Science:

• Non-destructive testing
• Advanced imaging techniques
• Signal processing

Background:

• Electromagnetic induction thermography (EIT) is vital for industrial defect detection, offering fast, contactless analysis.
• Limitations in hardware and complex test pieces pose challenges for current EIT systems.
• Weak defect signals are often obscured by noise and complex backgrounds in real-world scenarios.

Purpose of the Study:

• To develop an advanced video processing algorithm for defect detection in EIT.
• To address challenges posed by weak signals, noise, and complex backgrounds in EIT data.
• To improve the accuracy and efficiency of defect detection in non-destructive evaluation.

Main Methods:

• Proposed a novel hierarchical low-rank and sparse tensor decomposition method.
• Applied the algorithm to spatio-temporal video data from thermal imaging devices.
• Focused on mining anomalous patterns within the induction thermography stream.

Main Results:

• The proposed algorithm effectively suppresses background interference and sharpens defect visual features.
• It overcomes limitations of over- and under-sparseness found in existing algorithms.
• Real-time experiments demonstrated superior efficiency and accuracy compared to current methods.

Conclusions:

• The hierarchical low-rank and sparse tensor decomposition is a promising approach for EIT defect detection.
• This method enhances the capability to detect weak and sparse defect signals in noisy environments.
• The findings contribute to advancing non-destructive evaluation techniques in industrial applications.