Automatic Recognition of Worded and Diagrammatic Road Markings Based on Laser Reflectance Information

Abstract

Road markings play an important role in road safety and provide guidance as well as information to drivers. Most previous studies have focused on the detection and recognition of lane lines; however, more researchers have studied road markings of worded messages, symbols, and directional arrows recently. An algorithm for automatically detecting and recognizing worded and diagrammatic road markings on traffic lanes was studied and is presented in this paper. Road markings of Chinese characters are first studied. Moreover, a method for identifying the completeness of road markings is also presented. An approach was proposed to extract road marking features based on the derivation of a binary matrix image transformed from laser reflectance collected from a top-down roadway view laser scanner. Case studies show that the approach can detect and recognize road markings with a significant success rate. The success rate is 94.7% for Chinese character road markings and 96.2% for direction arrows and others. In this study, analyzed results indicate that a significant portion of road markings have relatively low completeness. It is suggested to have further investigation to determine the acceptance level and maintenance threshold of road marking completeness as future research work.