Urban Road Marker Classification Using Histogram of Oriented Gradient and Local Binary Pattern with Artificial Neural Network

Abstract

Road safety is paramount and depends on different factors, one of which is road makers. To regulate the road traffic while ensuring its safety, several common types of road markers are typically used such as the dashed, double solid and solid-dashed markers. This paper proposed a vision-based system to classify three types of markers using Artificial Neural Network (ANN) and image processing approaches. The accuracy was compared between two types of feature extraction techniques, namely the vector on Histogram of Oriented Gradient (HOG) and Local Binary Pattern (LBP), with the latter is lower at 98.3% vs 98.9%. The accuracy is further improved at 99.4% with both HOG and LBP features were combined and used as the input vectors for the classification process.