Abstract
Spatial resolution is a critical metric for evaluating image quality in eddy current imaging. In Carbon Fiber Reinforced Polymer (CFRP) inspection, undersampling often leads to spectral aliasing, causing fiber textures to appear coarse and blurred. This study investigates an eddy current image enhancement method based on spatial frequency domain de-aliasing to significantly improve spatial resolution and detection accuracy for fiber arrangements and defects. A mesoscopic electromagnetic model of CFRP laminates is established to reveal the imaging mechanism of fiber textures and the aliasing patterns of periodic structures. The Point Spread Function (PSF) of the eddy current probe is estimated, and the spatial resolution of different Transmitter-Receiver (T-R) probes is evaluated. Finally, an inverse convolution texture refinement focusing method based on PSF and Fourier transform is proposed to achieve high-precision imaging of fiber orientations and layup defects in CFRP laminates.
Introduction
Carbon Fiber Reinforced Polymer (CFRP) is widely used in aerospace and automotive industries for its high strength-to-weight ratio. However, non-destructive testing (NDT) of its internal structure remains challenging. Eddy Current Testing (ECT) is a popular method, but it faces a significant resolution gap: individual carbon fibers are approximately 7 μm in diameter, while sensing probes are typically several millimeters wide.
Because the probe is much larger than the fibers it scans, the resulting image suffers from spatial undersampling. In signal processing terms, this causes aliasing—high-frequency details of the fiber textures are folded into lower frequencies, resulting in a blurry, "coarse" image. This makes it difficult to identify critical meso-scale defects like fiber waviness, wrinkles, or small gaps.
Methodology
The authors address this blurring through a combination of simulation, hardware characterization, and digital signal processing.
Mesoscopic Simulation and T-R Probes
A detailed electromagnetic model was developed in COMSOL Multiphysics to simulate eddy current flow through individual fiber bundles. The study specifically utilized Transmitter-Receiver (T-R) probes, which were found to be more sensitive to fine fiber textures than traditional coaxial probes.
PSF Estimation
The system's response was characterized using the Point Spread Function (PSF). The PSF describes how a single point source is blurred by the imaging system. In this study, the PSF was estimated by scanning a test plate with a 1 mm diameter through-hole, representing a point-like defect. The response was modeled as a Gaussian function related to the probe's physical dimensions.
Inverse Convolution (De-aliasing)
The core innovation is a Fast Fourier Transform (FFT) deconvolution algorithm. By treating the physical size of the probe as a low-pass spatial filter, the "blur" becomes a predictable mathematical transformation. By taking the blurry raw image and "dividing out" the probe's PSF in the frequency domain, the fiber textures can be mathematically thinned and the image refocused.
Key Results
- Resolution Improvement: In experimental tests on a cross-ply laminate, the raw scan showed an average fiber texture width of 7.1 mm. After applying the de-aliasing algorithm, this was sharpened to 5.6 mm, a significant increase in spatial resolution.
- Visualizing Complex Layups: In complex three-direction layups ($[0^\circ/45^\circ/-45^\circ]_8$), where raw eddy current images were featureless, de-aliasing revealed the distinct orientations of the fiber layers.
- Defect Detection: The method successfully highlighted hidden delamination defects and "focused" their edges, allowing for more accurate sizing and shape identification of internal damage.
Conclusion
This research demonstrates that digital de-aliasing can achieve the resolution of a much smaller probe while maintaining the robust signal-to-noise ratio of a larger one. While the de-aliasing process can sometimes cause signals from fiber textures and defects to overlap, it represents a significant step toward high-precision automated inspection of composite materials. Future work will focus on signal decoupling to further separate structural textures from defect signals.
Summary updated