Detection and recognition of road markings for advanced driver assistance system

Abstract

This paper proposes a method for detecting direction indicators marked on road surfaces for safe driving support. In the proposed method, images are received from a vehicle’s black box, and a method for template matching is used on such direction indicators to detect the indicator area. By detecting the Maximally Stable Extremal Regions (MSER), the matching method is used to detect the road indicator area after the areas where road indicator candidate regions and binary code result images overlap are detected through the multi-level threshold template. The results of the experiment conducted in an actual vehicle driving environment show that, from the total of 270 frames that include indicators, each frame requires approximately 0.34 s, and a minimum of 83 % detection rate is provided.