Road detection based on the color space and cluster connecting

摘要

This paper propose a road detection, which is based on the color space and cluster connection. Generally, this method consists of two steps. The first step is road detection based on the color space. The second step is removing the fails road detection in the first step by using correlation. The correlation formula is used for calculating the similarity of each cluster image. The fails road detection will be removed, if their cluster image uncorrelated with neighbors cluster image. The simulation results show that our method has good performance which is capable significantly removing fails road detection. As well as evaluation by using precision, recall and accuracy have been presented.