Title :
Evaluating Color Representations for On-Line Road Detection
Author :
Alvarez, Jose M. ; Gevers, Theo ; Lopez, Antonio M.
Author_Institution :
NICTA, Canberra, ACT, Australia
Abstract :
Detecting traversable road areas ahead a moving vehicle is a key process for modern autonomous driving systems. Most existing algorithms use color to classify pixels as road or background. These algorithms reduce the effect of lighting variations and weather conditions by exploiting the discriminant/invariant properties of different color representations. However, up to date, no comparison between these representations have been conducted. Therefore, in this paper, we perform an evaluation of existing color representations for road detection. More specifically, we focus on color planes derived from RGB data and their most common combinations. The evaluation is done on a set of 7000 road images acquired using an on-board camera in different real-driving situations.
Keywords :
image classification; image colour analysis; image representation; image sensors; object detection; roads; traffic engineering computing; RGB data; color representation evaluation; color representations; discriminant-invariant properties; lighting variations; modern autonomous driving systems; moving vehicle; on-board camera; on-line road detection; pixel classification; real-driving situations; traversable road area detection; weather conditions; Cameras; Color; Image color analysis; Lighting; Roads; Robustness; Training; color; road detection; single class classifier;
Conference_Titel :
Computer Vision Workshops (ICCVW), 2013 IEEE International Conference on
Conference_Location :
Sydney, NSW
DOI :
10.1109/ICCVW.2013.82
