低对比度下利用激光网格的路面障碍识别方法  

Road obstacle recognition using laser mesh under low contrast

作　　者：任赵旭 姬书得[1] 林京鹏 胡为[1] 宋崎[1] REN Zhaoxu;JI Shude;LIN Jingpeng;HU Wei;SONG Qi(College of Aerospace,Shenyang Aerospace University,Shenyang 110136,China)

机构地区：[1]沈阳航空航天大学航空宇航学院

出　　处：《应用科技》2019年第4期21-25,共5页Applied Science and Technology

基　　金：辽宁省自然科学基金项目(20180550206);辽宁省教育厅科学研究项目(L201749)

摘　　要：为了解决目前视觉识别中低对比度图像识别速度低、精度差的问题,提出了一种基于红外激光网格路面障碍识别方法。该方法可以在低对比度条件下利用激光网格的错位作为障碍物的识别标志,得到障碍物的大小和距离。首先标定双目相机,并对图像坐标和三维空间坐标建立映射关系。之后利用道格拉斯普克算法进行网格提取,然后提取网格断点的三维坐标得到障碍物的大小和距离,最后对算法进行验证。结果表明该方法克服了视觉识别在特殊条件下识别效果不好的问题,得到了较为准确的障碍物轮廓和距离。In order to solve the problems of low speed and poor accuracy for low contrast image recognition in visual recognition,a road surface obstacle recognition method based on infrared laser mesh is proposed.In this method,the dislocation of laser mesh can be used as the recognition mark of obstacles under low contrast condition to obtain the size and distance of obstacles.Firstly,the binocular camera was calibrated,and the mapping relationship be?tween image coordinates and three?dimensional space coordinates was established.Then,the Douglas Poker algo?rithm was used to extract the grid,and then three?dimensional coordinates of grid breakpoints were extracted to get the size and distance of obstacles.Finally,the algorithm was verified by experiments.The experimental results showed that this method overcame the problem of poor recognition effect of visual recognition under special condi?tions,and obtained more accurate obstacle contours and distances.

关 键 词：视觉识别 低对比度 红外激光网格 错位 双目相机 网格提取 轮廓 距离 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]