Anchor-based characterization of knitted surface morphology with scale-defined ISO 25178 descriptors

Knitted fabrics exhibit hierarchical and scale-dependent surface morphology that is difficult to represent using a large set of partially overlapping areal descriptors. This study developed an anchor-based framework for the scale-defined characterization of knitted surface morphology and examined its association with air permeability. Nine finished fabrics representing three related knitted architectures were measured by focus-variation microscopy, providing 90 areal surface topography measurements. A common reference condition was established using an S-filter nesting index of 20 μm and an L-filter nesting index of 2.0 mm. Among 27 candidate descriptors defined within the ISO 25178 framework, correlation analysis identified 43 parameter pairs with |r|>0.90. Combining this covariation structure with physical interpretation produced a Characterization Parameter Set comprising the arithmetical mean height (Sa), autocorrelation length (Sal), and arithmetic mean peak curvature (Spc). These descriptors represent average vertical amplitude, lateral structural scale, and local peak geometry, respectively. Sa and Sal changed markedly with the L-filter condition, whereas Spc showed smaller mean changes; omission of the L-filter also altered the architecture-level ordering of Sal. Air permeability was subsequently examined using the nine fabric-condition means. A conventional benchmark based on areal mass and thickness showed the smallest internal leave-one-condition-out errors (Q 2 _LOOCV = 0.897; RMSE = 13.95 L m -2 s -1 ). The exploratory all-anchor CPS model achieved a close full-data fit (R 2 = 0.919) but showed larger leave-one-condition-out errors (Q 2 _LOOCV = 0.285; RMSE = 36.79 L m -2 s -1 ). The proposed framework provides a compact, physically interpretable, and scale-specified representation of knitted surface morphology for subsequent surface–performance investigations.