Using ground penetrating radar to classify features within structural timbers

Abstract

Recent developments in image processing play a significant role in identifying and classifying features on ground penetrating radar (GPR) radargrams. Two groups of timber specimens were examined in this study. The first group comprised laboratory prepared Douglas-fir (Pseudotsuga menziesii) timber sections with inserts of known internal characteristics. The second group comprised timber girders salvaged from timber bridges on historic Route 66 that had been exposed to weather for more than 80 years. A subsurface interface radar system with a 2-GHz palm antenna was used to scan these two groups of specimens. GPR sensed differences in dielectric constants along the scan path caused by the presence of water, metal, or air within the wood. This study focused on two procedures: feature identification and classification. Image processing was used to identify the presence of defect features, which was then classified based upon machine learning. GPR radargrams proved to be an excellent tool to detect and define characteristics within structural timbers.

Citation

Wu, XI; Senalik, Christopher Adam; Wacker, James; Wang, Xiping; Li, Guanghui. 2019. Using ground penetrating radar to classify features within structural timbers. In: Wang, X.; Sauter, U.H.; Ross, R.J., eds. 2019. Proceedings: 21st International Nondestructive Testing and Evaluation of Wood Symposium. General Technical Report FPL-GTR-272. Madison, WI: U.S. Department of Agriculture, Forest Service, Forest Products Laboratory. pp 502-510.