摘要
为解决传统SAD算法在基于半主动悬架的路面预瞄测距应用场景下匹配精度与测距精度较差的问题,首先对SAD视差窗口内中心像素周围像素点的权重重新分配,降低窗口边缘像素的影响;然后考虑到光照差异较大的场景引入Census变换,并采用子窗口灰度均值替代法增强其匹配鲁棒性,再将其与改进的SAD算法融合,最终形成融合CensusSAD算法。实验使用双目相机获取四种实际道路场景中的图像数据验证改进后的算法效果。实验表明融合CensusSAD算法的实时性有所下降但仍能满足要求,匹配精度比SAD算法高20%左右,在7m内融合Census-SAD算法测距精度比SAD算法精度要高3%左右。
In order to solve the problem that the traditionalSAD algorithm has poor matching accuracy and ranging accuracy in the application scene of road-preview ranging based on semi-active suspension,at first the weight of the pixels around the center pixel in the disparity window are redistributed so as to reduce the influence of the pixels at the edge of the window.And then considering the factors of large illumination difference in the real scenes,the Census algorithm is added.And the sub-window grayscale mean substitution method is used to enhance its matching robustness,and then it is fused with the improved SAD algorithm.Finally,the fusion Census-SAD algorithm has formed.The experiments use a binocular camera to acquire image data in four actual road scenes to verify the improved algorithm.The results show that the matching accuracy of the fusion Census-SAD algorithm is20%higher than SAD algorithm and the ranging accuracy of the fusion Census-SAD algorithm is about 3%better than SAD algorithm within 7 meters.
作者
官锌强
孟文
杨明亮
丁渭平
GUAN Xin-qiang;MENG Wen;YANG Ming-liang;DING Wei-ping(School of Mechanical Engineering,Southwest Jiaotong University,Sichuan Chengdu 610031,China;Engineering Research Center of Advanced Driving Energy-Saving Technology,Ministry of Education,Sichuan Chengdu 610031,China;Technology and Equipment of Rail Transit Operation and Maintenance Key Laboratory of Sichuan Province,Sichuan Chengdu 610031,China)
出处
《机械设计与制造》
北大核心
2022年第7期14-18,共5页
Machinery Design & Manufacture
基金
国家自然科学基金资助项目(51775451)
重庆市教委科学技术研究项目(KJQN201904009)。
