Mitigating spreading effects in coplanar capacitive sensing via regularized image deconvolution using an estimated PSF

Accurate defect sizing is critical for advancing coplanar capacitive sensing toward proactive safety management applications such as structural health monitoring and damage prognosis. However, the inherent spreading effect of the coplanar capacitive sensors (CCSs) hampers precise defect characterization, limiting the reliability of capacitive imaging (CI) in such contexts. To address this, we applied image deconvolution and combined finite element (FE) simulations and experiments to evaluate and compare two regularization-based algorithms—L2 (Tikhonov) and Total Variation (TV)—for CI applications. Using estimated PSFs and optimized regularization parameters, we assessed their effectiveness: results show that both methods reduced spreading, with TV demonstrating superior noise suppression and boundary restoration, particularly in conducting specimens. The spreading effect was found to be material- and direction-dependent, exhibiting greater asymmetry in conducting specimens. Automatically selected regularization parameters (λ) followed consistent trends: experimental data required stronger regularization, while conductive samples improved λ stability. TV consistently yielded the narrowest full-width-at-zero (FWZ) ranges, although only modestly outperforming L2. Overall, both optimized L2 and TV regularization-based deconvolution approaches effectively mitigate anisotropic spreading in CCS image data, yielding defect‐size estimates closely matching actual dimensions and strengthening the coplanar capacitive sensing technique for proactive safety management applications. However, while PSF estimation proved beneficial, further refinement in its modeling/estimating is needed to improve deconvolution accuracy.