A width centric image to mechanics framework for assessing advance grouting with TMSE adaptive width segmentation and lab scale validation

Grout veins formed by advance grouting influence bulk stiffness and permeability. We present a width‑centric, image‑to‑mechanics framework that performs SES preprocessing (HDMR‑based multi‑feature fusion and a moment‑based sub‑pixel localizer), extracts graph‑theoretic centerlines, and estimates vein widths via TMSE-adaptive segmentation, which directly supports property estimation. On a lab‑scale image set, SES improves segmentation across seven backbones, and the resulting width/orientation statistics yield laboratory‑scale estimates of elastic modulus and permeability. Worst‑case errors are 8.83% (modulus) and 10.42% (permeability); these do not meet design‑grade tolerances. The method is therefore positioned for screening/monitoring under controlled conditions, not for design decisions. We clarify assumptions (local monotonic edge profile, 2‑D effective analysis), provide stopping criteria and parameters for reproducibility.