基于单目视觉的道路图像理解综述

An Overview of Road Image Understanding Based on Monocular Vision

基于单目视觉的环境感知一直是机器视觉领域的重要研究内容。视觉信息处理技术主要应用于道路检测识别、路标检测识别、障碍检测识别、障碍跟踪及其他传感器的自标定、辅助标定等几个主要方面。道路检测识别是环境感知、场景理解中的关键技术之一。道路图像理解不仅能够从道路图像中理解道路环境、获取未知信息,而且能为其他任务(如行人和车辆检测、路径规划等)提供线索。但同时,基于视觉的道路图像理解也是一项极具挑战性的任务,因为道路场景的检测易受光照、天气、遮挡等因素的影响。

The ambient awareness based on monocular vision is always the important research in the computer vision. The visualinformation processing almost has been applied into the some aspects, such as road detection and identification, sign detection andidentification, block detection and identification, block tracking and other sensors demarcated and aided demarcated. Road imageunderstanding not only can master road environment and fetch unknown information from road image, but also can provide intelligence for other missions as pedestrian, vehicle detection and route planning, etc. However, road image understanding based on vision is also a challenging mission, because the road scene detection vulnerable to light, weather and shelter etc.