Nearest Neighbor Classifier for Vision Based Unstructured Road Detection

Driver can drive a vehicle on road very smoothly, if he knows about the road and non road region in surrounding environments. Accurate road detection is an important and difficult task for the development of vision based autonomous land vehicle. Therefore this research work proposed an algorithm which can separate road and non road region effectively. Nearest neighbor algorithm has trained with chromatic information to differentiate the road and surrounding environments. Created classifier divide the image of road region into three classes such as road region, near background region and far background region. Classifier training process utilized chromatic information from that color space which has more resemblance with human color perception. System performance is evaluated on the basis of precision, recall and required training time. Achieved results signify that overall performance of the system is worthy.