Current path transformation models for log knot detection based on electric resistance tomography using path distortion compensation
In electrical resistance tomography (ERT) of logs, soft-field current deflection caused by high-impedance knots distorts electrical signals, yet systematic comparisons of current path models remain limited. This study compared four current path models (elliptical, rectangular, along-bark edge, and bark-to-knot composite) using 11 natural-knot logs (8 for training and 3 for independent validation) and developed an MLR-based distortion compensation framework. The framework transformed resistance measurements from any assumed path into elliptical-equivalent values for eccentric circle-based inverse-distance-weighted (ECIDW) reconstruction. The elliptical path achieved the lowest average area prediction error (18.61%). MLR compensation reduced the composite-path area prediction error from 79.94–96.50% to 0.79–27.16%. Leave-one-log-out cross-validation demonstrated excellent generalization for the rectangular path model (ΔR² = 0.030), whereas the along-bark and composite models showed larger generalization gaps (ΔR² = 0.44–0.51), mainly due to limited training sample diversity. For the along-bark path, compensation did not improve validation accuracy on average (26.63% vs. 35.64%). These results demonstrate the effectiveness of the framework for path geometries whose distortion is systematic, and support its application to larger datasets and diverse wood species.
Authors
- Fenglu Liu (ORCID: https://orcid.org/0009-0008-5219-3680)
- Jiawei Xiao (ORCID: https://orcid.org/0009-0002-5802-5625)
- Ronghui Xu
- Shengyu Lin
- Yizhuan Li
- Guoqiang Lin
Institutions
- Wuyi University (CN)
- Fujian Agriculture and Forestry University (CN)
- Wuyi University (CN)
Publication Details
- Journal
- Wood Material Science and Engineering
- Published
- 2026-11-20
- DOI
- https://doi.org/10.1080/17480272.2026.2763647
- Primary Topic
- Electrical and Bioimpedance Tomography
- Type
- article
- Field-Weighted Citation Impact
- 0.00
Funders
- National Natural Science Foundation of China
- Wuyi University