The high-density resistivity method, as an effective technique for detecting subsurface electrical structures, shows significant potential in identifying concealed defects such as voids and water-rich zones beneath highway subgrades. This study constructs stratified models of highway subgrades featuring various defect types and employs the Gauss-Newton inversion method to explore the electrical characteristics of these defects. This provides a theoretical foundation for the application of high-density electrical methods in the precise detection and identification of subgrade defects. To further delineate the position and morphology of anomalies, the fuzzy C-means clustering algorithm is applied to the inversion data. The results indicate that using clustering analysis endows the inversion outcomes with vertical stratification characteristics, more clearly reflecting the resistivity differences caused by different fillings within the subgrade structure model. This method, leveraging the depth advantages of high-density electrical sounding, enhances the delineation of the boundaries of deep-seated defects within highway subgrades and more accurately identifies the distribution of anomalies. It provides technical support for the detection and prevention of highway subgrade defects.