Driving space detection for structured outdoor environment using monocular vision

摘要

Driving space detection is one of the basic steps for autonomous guidance of vehicles. In this paper a real time road segmentation algorithm is proposed for identifying driving space. The presented algorithm uses the concept of appearance based object detection (ABOD) to detect traversable area for vehicles. The input image is partitioned as road and off road and each individual pixel is classified as either traversable or non traversable. It uses Mean Shift Algorithm for road segmentation and then uses ABOD on segmented image to separate road and off road areas.