测绘通报 ›› 2019, Vol. 0 ›› Issue (12): 22-25.doi: 10.13474/j.cnki.11-2246.2019.0379

• 人工智能与高精度地图 • 上一篇    下一篇

基于高分辨率遥感影像的车道线提取

张世强, 王贵山   

  1. 北京灵图软件技术有限公司, 北京 100193
  • 收稿日期:2019-03-01 发布日期:2020-01-03
  • 作者简介:张世强(1971-),男,工程师,主要从事电子地图、遥感影像处理等工作。E-mail:geomap@sohu.com

Extraction of lane line from high resolution remote sensing image

ZHANG Shiqiang, WANG Guishan   

  1. Beijing Lingtu Software Co., Ltd., Beijing 100193, China
  • Received:2019-03-01 Published:2020-01-03

摘要: 车道线是自动驾驶高精度地图的组成部分,利用车载摄像头和激光雷达提取车道线的研究已经比较深入。本文探讨了利用高分辨率遥感影像提取车道线的方法。首先依据直方图原则选取合适的阈值进行二值化分割,其次利用道路中线形成的缓冲区去除道路范围外的要素,然后利用形态学算子去除噪声,最后利用方向和面积因子提取不同的车道线。该方法可以降低数据采集的成本,提高采集效率。

关键词: 高精地图, 车道线, 高分辨率, 遥感影像, 自动驾驶

Abstract: Lane line is an important part of the high definition(HD) map. This paper discusses the method of lane line extraction using high resolution remote sensing image. According to the histogram principle, the appropriate threshold is selected for binarization segmentation, the buffer formed by the road center line is used to remove the elements outside the range of the road, and the morphological operator is used to remove the noise. Shape and direction feature used to distinction lane line. This method can reduce the cost of data collection and improve the efficiency of data collection.

Key words: high definition(HD) map, lane line, high resolution, remote sensing image, self-driving

中图分类号: