摘要
针对当前基于双目视觉的道路环境分析实时性差、检测不准确等问题,提出了一种改进V视差法的道路区域检测算法。该算法首先对原始图片进行车道线检测确定道路消失点,从而确定图像的感兴趣区域。然后,使用极大最小值约束获取V视差图中的斜线,从而提取道路区域。实验结果表明,由于该方法在确定感兴趣区域后计算原始视差图,因此,速度提高了29.71%,且相对于传统V视差法,算法更好地实现了路面分割;同时,障碍物检测的精确率和召回率两个指标分别提高了2.165%和4.837%。基于该算法具有良好的准确性和实时性,能有效识别道路中的障碍物,因此,可以为车辆提供可行驶区域以及为驾驶员提供辅助作用。
Aiming at the problems of poor real-time performance and inaccurate detection of road environment analysis based on binocular vision, a road area detection algorithm based on improved V-disparity method is proposed. The algorithm firstly detects the lane line of the original image to determine the vanishing point of the road, thereby determining the region of interest of the image. Then, the oblique line in the V-disparity map is obtained using the maximum and minimum constraints, so as to extract the road area. Experimental results show that because the method calculates the original disparity map after determining the region of interest, the speed is increased by 29.71%, and compared with the traditional V-disparity method, the algorithm achieves better road segmentation;at the same time, the accuracy of obstacle detection The two indicators of and recall rate increased by 2.165% and 4.837% respectively. Based on this algorithm, it has good accuracy and real-time performance, and can effectively identify obstacles in the road, so it can provide a drivable area for the vehicle and provide assistance to the driver.
作者
李春明
耿永鹏
远松灵
LI Chunming;GENG Yongpeng;YUAN Songling(School of Information Science and Engineering,Hebei University of Science and Technology,Shijiazhuang 050018,China;Shjiazhuang Jinghua Electronic Industrial Co.,Ltd.,Shijiazhuang 050200,china)
出处
《激光杂志》
CAS
北大核心
2022年第1期107-112,共6页
Laser Journal
基金
河北省科技厅项目(No.17210803D)。
关键词
双目视觉
辅助驾驶
道路检测
感兴趣区域
V视差图
binocular vision
assist driving
road detection
region of interest
V disparity map
