摘要
针对夜间获取的车道图像对比度低的问题,提出了一种基于自适应阈值分割的夜间车道标识线识别算法。首先,对预处理后的图像进行分块拉伸以增强边缘信息。再结合Otsu门限法和邻域中值法,通过加权分配获取自适应阈值,对车道图像进行分割。最后,采用分区域搜索方式,利用Hough变换精确地提取车道标识线。现场实测表明,针对结构化道路的车道线,论文采用的车道线提取方法准确率高且实时性好。
For the low contrast of the lane images obtained from the night ,a nighttime lane markings recognition based on adaptive threshold segmentation is proposed .The pre‐processing images are first blocked and stretched for edge enhance‐ment .An adaptive threshold ,determined by Otsu’ thresholding and neighborhood median method ,is used to segment lane images .The lane markings are finally extracted by using Hough transform via sub‐region searching .The estimation shows that the proposed lane recognition has high accuracy and good real‐time processing for structured road extraction .
出处
《计算机与数字工程》
2015年第2期308-311,共4页
Computer & Digital Engineering
基金
中国民航大学中国民航信息技术科研基地开放课题基金项目(编号:CAAC-ITRB-201205)资助
