基于改进Canny算法的道路标线自动识别及定位
|
摘要
在自动化道路病害检测过程中,检测车的横向偏移会导致路面病害横向定位的误差,通过边缘检测算法对标线进行自动识别及定位,可消除病害横向定位的误差.但传统Canny边缘检测算法存在对标线边缘识别不全,噪声、伪边缘较多的问题.因此本文对Canny算法进行改进:采用维纳滤波替代高斯滤波以滤除噪声;利用OTSU法自适应地选取图像阈值;提出划定准偏区域以滤除路面干扰并提高边缘完整识别率.研究结果表明:与其他边缘检测算法相比,该算法检测出的边缘像素总数分布在900～2 000,与标准值1 352较为符合,8连通域数与边缘像素总数、4连通域数的比值分别分布在0～0.04和0～0.29,在抑制噪声、剔除伪边缘及路面干扰、边缘识别连续性和单一边缘响应方面实际应用效果较好;人工法验证结果表明该算法定位标线结果与实际值的吻合度高达99%,可在同步采集的3D激光图像上进行路面病害横向定位.
In a pavement distress survey, a wandering survey vehicle would reduce the accuracy of lateral position measurement of pavement distress. The errors in the measurement of distress lateral position can be eliminated based on the lane marking identified using an edge detection algorithm. However,the traditional edge detection algorithm has some limitations,such as incomplete and false edge recognition. Therefore,an improved Canny edge detection algorithm is proposed herein,which is composed of three steps:noise elimination using a Wiener filter,adaptive threshold definition using Otsu's method,and lane marking edge recognition based on offset-allowed area method. The test results show that the total number of edge pixels detected by the proposed algorithm is between 900 and 2 000,which is close to the standard value of 1 352 for existing edge detection algorithms. The ratio of the number of eight-connected domains to the total number of edge pixels ranges from 0 to 0.04,and the ratio of the number of eight-connected domains to the number of four-connected domains ranges from 0 to 0.29,which shows that the improved algorithm demonstrates excellent performance in terms of noise elimination,false edge recognition,consistent edge detection,and single edge response. A validation test based on the manual method shows that the correlation between the test value and reference is up to 99%. The proposed method can be used to detect the lateral position of pavement distress based on 3 D laser imaging data.
In a pavement distress survey, a wandering survey vehicle would reduce the accuracy of lateral position measurement of pavement distress. The errors in the measurement of distress lateral position can be eliminated based on the lane marking identified using an edge detection algorithm. However,the traditional edge detection algorithm has some limitations,such as incomplete and false edge recognition. Therefore,an improved Canny edge detection algorithm is proposed herein,which is composed of three steps:noise elimination using a Wiener filter,adaptive threshold definition using Otsu's method,and lane marking edge recognition based on offset-allowed area method. The test results show that the total number of edge pixels detected by the proposed algorithm is between 900 and 2 000,which is close to the standard value of 1 352 for existing edge detection algorithms. The ratio of the number of eight-connected domains to the total number of edge pixels ranges from 0 to 0.04,and the ratio of the number of eight-connected domains to the number of four-connected domains ranges from 0 to 0.29,which shows that the improved algorithm demonstrates excellent performance in terms of noise elimination,false edge recognition,consistent edge detection,and single edge response. A validation test based on the manual method shows that the correlation between the test value and reference is up to 99%. The proposed method can be used to detect the lateral position of pavement distress based on 3 D laser imaging data.
引文
[1]WANG Y,TEOH E K,SHEN D.Lane detection and tracking using B-snake[J].Image&Vision Computing,2004,22(4):269-280.
[2]LI Zuoquan,MA Huimin,LIU Zhengyu.Road lane detection with Gabor filters[C]//International Conference on Information System and Artificial Intelligence.[S.l.]:IEEE,2016:436-440.
[3]WANG Junguo,LIN Chengjian,CHEN Shyiming.Applying fuzzy method to vision-based lane detection and departure warning system[J].Expert Systems with Applications,2010,37(1):113-126.
[4]CAO Jianchao,LIN Qiuhua,FU Mengyin.Extraction of single-pixel-width single edge using dynamic thresholding for lane detection on structured roads[C]//IEEE China Summit&International Conference on Signal&Information Processing.[S.l.]:IEEE,2014:460-464.
[5]王智文.几种边缘检测算子的性能比较研究[J].制造业自动化,2012,34(11):14-16.WANG Zhiwen.Comparison research of capability of several detection operators for edge detection[J].Manufacturing Automation,2012,34(11):14-16.
[6]赵芳,栾晓明,孙越.数字图像几种边缘检测算子检测比较分析[J].自动化技术与应用,2009,28(3):68-69.ZHAO Fang,LUAN Xiaoming,SUN Yue.Edge detection operators in digital image processing[J].Techniques of Automation and Applications,2009,28(3):68-69.
[7]CANNY J.A computational approach to edge detection[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,1986,8(6):679-698.
[8]张斌,贺赛先.基于Canny算子的边缘提取改善方法[J].红外技术,2006,28(3):165-169.ZHANG Bin,HE Saixian.Improved edge-detection method based on Canny algorithm[J].Infrared Technology,2006,28(3):165-169.
[9]BAO P,ZHANG Lei,WU Xiaolin.Canny edge detection enhancement by scale multiplication[J].IEEETransactions on Pattern Analysis and Machine Intelligence,2005,27(9):1485-1490.
[10]ZHAO Huili,QIN Guofeng,WANG Xingjjian.Improvement of Canny algorithm based on pavement edge detection[C]//International Congress on Image&Signal Processing.[S.l.]:IEEE,2010,2:964-967.
[11]许宏科,秦严严,陈会茹.一种基于改进Canny的边缘检测算法[J].红外技术,2014,36(3):210-214.XU Hongke,QIN Yanyan,CHEN Huiru.An improved algorithm for edge detection based on Canny[J].Infrared Technology,2014,36(3):210-214.
[12]苗锡奎,胡启立,赵威,等.一种新的图像边缘检测算法研究[J].光电子技术,2016,36(1):49-54.MIAO Xikui,HU Qili,ZHAO Wei,et al.The research of a new image edge detection method[J].Optoelectronic Technology,2016,36(1):49-54.
[13]吴翔,于微波,马艳辉,等.一种新的改进Canny图像边缘检测算法[J].影像科学与光化学,2016,34(1):116-121.WU Xiang,YU Weibo,MA Yanhui,et al.A new improved Canny image edge detection algorithm[J].Imaging Science and Photochemistry,2016,34(1):116-121.
[14]凌凤彩,康牧,林晓.改进的Canny边缘检测算法[J].计算机科学,2016,43(8):309-312.LING Fengcai,KANG Mu,LIN Xiao.Improved Canny edge detection algorithm[J].Computer Science,2016,43(8):309-312.
[15]WANG K C P,HOU Zhiqiong,GONG Weiguo.Automated road sign inventory system based on stereo vision and tracking[J].Computer-Aided Civil and Infrastructure Engineering,2010,25(6):468-477.
[16]彭博,WANG K C P,陈成,等.基于各向异性测度的路面三维图像裂缝识别[J].西南交通大学学报,2014,49(5):888-895.PENG Bo,WANG K C P,CHEN Cheng,et al.3Dpavement crack image detection based on anisotropy measure[J].Journal of Southwest Jiaotong University,2014,49(5):888-895.
[17]LUO Wenting,WANG K C P,LI Lin.Surface drainage evaluation for rigid pavements using IMU and1mm 3D texture data[J].Transportation Research Record:Journal of the Transportation Research Board,Pavement Management,2014,3:121-128.
[18]LUO Wenting,WANG K C P,LI Lin.Hydroplaning on sloping pavements based on inertial measurement unit(IMU)and 1 mm 3D laser imaging data[J].Periodica Polytechnica Transportation Engineering,2016,44(1):42-49.
[19]OTSU N.A threshold selection method from gray-level histograms[J].IEEE Transactions on Systems,Man,and Cybernetics,1979,9(1):62-66.
[20]林卉,赵长胜,舒宁.基于Canny算子的边缘检测及评价[J].黑龙江工程学院学报,2003,17(2):3-6.LIN Hui,ZHAO Changsheng,SHU Ning.Edge detection based on Canny operator and evaluation[J].Journal of Heilongjiang Institute of Technology,2003,17(2):3-6.
[2]LI Zuoquan,MA Huimin,LIU Zhengyu.Road lane detection with Gabor filters[C]//International Conference on Information System and Artificial Intelligence.[S.l.]:IEEE,2016:436-440.
[3]WANG Junguo,LIN Chengjian,CHEN Shyiming.Applying fuzzy method to vision-based lane detection and departure warning system[J].Expert Systems with Applications,2010,37(1):113-126.
[4]CAO Jianchao,LIN Qiuhua,FU Mengyin.Extraction of single-pixel-width single edge using dynamic thresholding for lane detection on structured roads[C]//IEEE China Summit&International Conference on Signal&Information Processing.[S.l.]:IEEE,2014:460-464.
[5]王智文.几种边缘检测算子的性能比较研究[J].制造业自动化,2012,34(11):14-16.WANG Zhiwen.Comparison research of capability of several detection operators for edge detection[J].Manufacturing Automation,2012,34(11):14-16.
[6]赵芳,栾晓明,孙越.数字图像几种边缘检测算子检测比较分析[J].自动化技术与应用,2009,28(3):68-69.ZHAO Fang,LUAN Xiaoming,SUN Yue.Edge detection operators in digital image processing[J].Techniques of Automation and Applications,2009,28(3):68-69.
[7]CANNY J.A computational approach to edge detection[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,1986,8(6):679-698.
[8]张斌,贺赛先.基于Canny算子的边缘提取改善方法[J].红外技术,2006,28(3):165-169.ZHANG Bin,HE Saixian.Improved edge-detection method based on Canny algorithm[J].Infrared Technology,2006,28(3):165-169.
[9]BAO P,ZHANG Lei,WU Xiaolin.Canny edge detection enhancement by scale multiplication[J].IEEETransactions on Pattern Analysis and Machine Intelligence,2005,27(9):1485-1490.
[10]ZHAO Huili,QIN Guofeng,WANG Xingjjian.Improvement of Canny algorithm based on pavement edge detection[C]//International Congress on Image&Signal Processing.[S.l.]:IEEE,2010,2:964-967.
[11]许宏科,秦严严,陈会茹.一种基于改进Canny的边缘检测算法[J].红外技术,2014,36(3):210-214.XU Hongke,QIN Yanyan,CHEN Huiru.An improved algorithm for edge detection based on Canny[J].Infrared Technology,2014,36(3):210-214.
[12]苗锡奎,胡启立,赵威,等.一种新的图像边缘检测算法研究[J].光电子技术,2016,36(1):49-54.MIAO Xikui,HU Qili,ZHAO Wei,et al.The research of a new image edge detection method[J].Optoelectronic Technology,2016,36(1):49-54.
[13]吴翔,于微波,马艳辉,等.一种新的改进Canny图像边缘检测算法[J].影像科学与光化学,2016,34(1):116-121.WU Xiang,YU Weibo,MA Yanhui,et al.A new improved Canny image edge detection algorithm[J].Imaging Science and Photochemistry,2016,34(1):116-121.
[14]凌凤彩,康牧,林晓.改进的Canny边缘检测算法[J].计算机科学,2016,43(8):309-312.LING Fengcai,KANG Mu,LIN Xiao.Improved Canny edge detection algorithm[J].Computer Science,2016,43(8):309-312.
[15]WANG K C P,HOU Zhiqiong,GONG Weiguo.Automated road sign inventory system based on stereo vision and tracking[J].Computer-Aided Civil and Infrastructure Engineering,2010,25(6):468-477.
[16]彭博,WANG K C P,陈成,等.基于各向异性测度的路面三维图像裂缝识别[J].西南交通大学学报,2014,49(5):888-895.PENG Bo,WANG K C P,CHEN Cheng,et al.3Dpavement crack image detection based on anisotropy measure[J].Journal of Southwest Jiaotong University,2014,49(5):888-895.
[17]LUO Wenting,WANG K C P,LI Lin.Surface drainage evaluation for rigid pavements using IMU and1mm 3D texture data[J].Transportation Research Record:Journal of the Transportation Research Board,Pavement Management,2014,3:121-128.
[18]LUO Wenting,WANG K C P,LI Lin.Hydroplaning on sloping pavements based on inertial measurement unit(IMU)and 1 mm 3D laser imaging data[J].Periodica Polytechnica Transportation Engineering,2016,44(1):42-49.
[19]OTSU N.A threshold selection method from gray-level histograms[J].IEEE Transactions on Systems,Man,and Cybernetics,1979,9(1):62-66.
[20]林卉,赵长胜,舒宁.基于Canny算子的边缘检测及评价[J].黑龙江工程学院学报,2003,17(2):3-6.LIN Hui,ZHAO Changsheng,SHU Ning.Edge detection based on Canny operator and evaluation[J].Journal of Heilongjiang Institute of Technology,2003,17(2):3-6.