Detection of night-time road lines captured from a moving platform for the purpose of retro-reflective quality assessment

A robust method for the extraction of road lines from an image is proposed. The images, captured from a moving platform, are of a night-time road scene. All images are tagged with synchronised global positioning (GPS) data. The images, once captured, are then stored and processed offline. The road lines are detected using a support vector machine (SVM). It is proposed that a photometric analysis is then carried out on the extracted line data and a database of effective and defective road lines and their locations is compiled.