Evaluation of unwanted road marking crossing detection using real-traffic data for intelligent transportation systems

Abstract

The authors focus in the contribution on finding and testing effective digital image processing methods in line crossing detection systems. To verify experimental real-world results the authors used the Matlab 2013a and Digital Signal Processing Toolbox software tools. In the line detection block in the designed application the driving lane is searched using segmentation techniques suitable for line detection. The information on road position (driving lanes) has been acquired based on edge detection of guiding lines. For line detection, the proposed modifications of Hough transform and local processing methods have been used. During the testing process testing video-sequences were created with several parameters: traffic situations (day-drive, night-drive, drive on distinct road types, highway drive etc.), temporary traffic signs (presence of temporary traffic signs in form of orange lines, absence of temporary traffic signs). A change of light conditions has also been simulated during testing by adding an additional noise and by determining its effects on line detection.