Advanced runway groove identification

Abstract

The automatic groove identification technique for runway and highway pavements is relatively new in pavement engineering research although many algorithms have been developed and proven efficient and successful in the last decade. However, the low-accuracy results of groove identification for deteriorated pavements remain prevalent. This study develops a software called NTUGroove which significantly improve the accuracy and efficiency of the groove identification under both good and deteriorated conditions. The developed algorithm includes Butterworth filter application, preliminary groove identification, geometric dimension calculation, false groove noise elimination, and mis-elimination compensation. These steps aim to include all potential grooves, eliminate the likely noises, and save the possibly misjudged eliminations. Four test sections with concrete and asphalt pavement provided by the US Federal Aviation Administration (FAA) and Taiwan Taoyuan International Airport are used in the case studies to compare the performances of FAA ProGroove (with default or with threshold) and NTUGroove. Grooves in good or deteriorated conditions can be identified by NTUGroove with a mean error rate of 1.7%, and the corresponding geometric dimensions are analyzed. The precision and recall values of the confusion matrix are used for program comparison. The mean precision and mean recall are 99.7% and 99.2%, respectively. NTUGroove is highly effective in groove identification and provides promising results for groove quantity counting and precision and recall values, even for seriously deteriorated sections.