期刊文献+

基于3D特征的道路检测算法 被引量:1

Road detection algorithm based on 3Dfeatures
下载PDF
导出
摘要 随着机动车辆保有量迅速攀升,给社会的安全问题带来巨大挑战,先进驾驶辅助系统(ADAS)越来越受到人们的重视,道路检测算法是ADAS的关键技术之一。针对传统的基于V视差图和法向量图的道路检测算法出现的问题,提出了一种基于3D特征的道路检测算法。利用双目视觉进行道路信息采集,通过立体匹配算法获得视差图,并结合相机内参对图像进行三维重建,然后计算像素的两种3D特征,得到道路场景的3D特征图。研究基于隐马尔可夫模型(HMM)的道路检测算法,针对其在拥堵场景分割效果不佳及容易被道路以外平坦区域干扰的缺点,提出改进算法。在KITTI数据集上对两种算法进行对比,提出的改进算法可以更精确的在不同的道路场景中检测出道路区域。 With the rapid increase of the number of motor vehicles,it brings a great challenge to the social security.Advanced driving assistance system(ADAS)has been paid more and more attention.Road detection algorithm is one of the key technologies of ADAS.Aiming at the problems of traditional road detection algorithms based on V-disparity map and normal vector map,a 3Dfeature-based road detection algorithm is proposed.Using binocular vision to collect road information,using stereo matching algorithm to obtain disparity map,and combining camera internal parameters to 3Dreconstruction of the image,and then calculating two kinds of 3Dfeatures of pixels to get 3Dfeature map of road scene.This paper studies the road detection algorithm based on Hidden Markov model(HMM).Aiming at the disadvantages of poor segmentation effect in congested scene and easy to be disturbed by flat areas outside the road,an improved algorithm is proposed.Comparing the two algorithms on the KITTI dataset,the improved algorithm proposed can detect the road area more accurately in different road scenes.
作者 童浩 Tong Hao(School of Computer and Information,HoHai University,Nanjing 211100,China)
出处 《电子测量技术》 2020年第10期106-112,共7页 Electronic Measurement Technology
关键词 道路检测 双目视觉 立体匹配算法 3D特征 隐马尔可夫模型 road detection binocular vision stereo matching algorithm 3Dfeature hidden Markov model
  • 相关文献

同被引文献2

引证文献1

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部