Robotics and Computer Vision Laboratory Login  
  Robotics and Computer Vision Laboratory kaist logo
Archive Courses

Publications
Home  >  Research  >  Publications

 
[International Conference] Robust Road Detection Based On Optimal Fusion Ratio of Classifier
2009 International Conference on Mechatronics and Information Technology (ICMIT) , November 2009
Download
  ICMIT09_ROBUST ICMIT09_ROBUST ROAD DETECTION BASED ON OPTIMAL FUSION RATIO OF CLASSIFIER.pdf (175.2K) [133]
Abstract

Road detection is an important issue for outdoor unmanned vehicle navigation. However, it is difficult to determine because outdoor exist many kinds of materials and environment changes such as illumination and weather. In this paper, we present a method of road detection based on optimal classifier fusion. We extract a ground plane region using geometric information that is robust to the environmental changes. In this region, a suitable area for driving is determined based on photometry classifiers, hue and texture. For accurate classification, we determine optimal fusion ratio through recognition rate and correct classification rate for each ratio of classifiers. The ratio gives two different results according to the object of the material classes. Experimental results show that our approach can detect the road effectively in different environments.
 
Key words - Road detection, optimal classifier fusion, optimal fusion ratio.


 
   
 

Robotics and Computer Vision Laboratory
KAIST | Electrical Engineering | Contact Us | Sitemap