基于改进Canny算子的铁路扣件定位方法
|
摘要
边缘特征是铁路扣件定位及缺陷分类的重要依据.针对传统Canny算子人为设定高低阈值会影响边缘检测效果,以及对多余信息消除能力较弱等缺点,提出一种轨道扣件边缘定位的改进Canny算子算法.首先,改变梯度幅值计算,除考虑水平和垂直两方向外,再增加计算45°与135°方向的梯度幅值,以有效避免边缘间断;其次,采用二维最大类间方差算法确定最优阈值以消除伪边缘,并以此进行Canny算子优化.仿真表明,该方法有效优化了Canny算子,能够准确检测所确定目标的真实边缘,在光照不均匀条件下的效果尤其突出,且具备较强的自适应性与抗噪能力,可基本消除因伪边缘而影响Hough变换的直线提取性能,定位准确率可达99.3%.
Edge feature is an important basis for positioning method and defect classification of railway fasteners.To overcome these shortcomings such as the artificial setting of the high and low thresholds in edge detection,and weak elimination of irrelevant information by Canny algorithm,a method of edge detection based on improved Canny algorithm is proposed.In order to optimize the Canny algorithm,firstly,the gradient amplitude calculation has been changed.Based on the horizontal direction and vertical direction,gradient templates are increased by 45 degrees and 135 degrees directions calculation of the gradient amplitude.This method can effectively avoid the discontinuous edges.Secondly,the thresholds are obtained by using the 2 D maximum between-cluster variance method instead of artificial thresholds which determines the optimal threshold to avoid false edges.Simulation experiments show that this method can effectively optimize the Canny algorithm and accurately detect the real edges.The method proposed has a remarkable effect in the case of uneven illumination,and has strong adaptability and anti-noise property.The line extraction performance of Hough transform can be basically eliminated due to the irrelevant false edge and the positioning accuracy can reach 99.3%.
Edge feature is an important basis for positioning method and defect classification of railway fasteners.To overcome these shortcomings such as the artificial setting of the high and low thresholds in edge detection,and weak elimination of irrelevant information by Canny algorithm,a method of edge detection based on improved Canny algorithm is proposed.In order to optimize the Canny algorithm,firstly,the gradient amplitude calculation has been changed.Based on the horizontal direction and vertical direction,gradient templates are increased by 45 degrees and 135 degrees directions calculation of the gradient amplitude.This method can effectively avoid the discontinuous edges.Secondly,the thresholds are obtained by using the 2 D maximum between-cluster variance method instead of artificial thresholds which determines the optimal threshold to avoid false edges.Simulation experiments show that this method can effectively optimize the Canny algorithm and accurately detect the real edges.The method proposed has a remarkable effect in the case of uneven illumination,and has strong adaptability and anti-noise property.The line extraction performance of Hough transform can be basically eliminated due to the irrelevant false edge and the positioning accuracy can reach 99.3%.
引文
[1] 闵永智,肖本郁,党建武,等.轨道扣件缺失的机器视觉快速检测方法[J].上海交通大学学报,2017(10):1268-1272.
[2] 陈金胜,陈兴杰,彭乐乐,等.基于边缘特征的无砟轨道扣件定位方法[J].计算机测量与控制,2018,26(5):172-175.
[3] LI X,ZHANG H.An improved canny edge detection algorithm[C]//2017 8th IEEE International Conference on Software Engineering and Service Science.Beijing:IEEE Press,2017:24-26.
[4] 周咏,顾桂梅,冉建民.基于改进主动轮廓模型的钢轨表面缺陷分割[J].兰州交通大学学报,2018,37(3):55-63.
[5] XIA Y Q,XIE F Y,JIANG Z Q.Broken railway fastener detection based on adaboost algorithm[C]//International Conference on Optoelectronics and Image Processing.New York:IEEE Press,2010:313-316.
[6] 张洋.巡道车视频处理系统关键技术的研究[D].北京:北京交通大学,2008.
[7] 范宏.基于图像的铁路扣件缺陷检测算法研究[D].成都:西南交通大学,2012.
[8] 万超.轨道扣件图像定位和特征识别算法研究[D].南昌:南昌大学,2015.
[9] 吴禄慎,万超,陈华伟,等.一种改进的十字交叉轨道扣件定位方法[J].铁道标准设计,2016(12):49-53.
[10] 何华武.建立中国高速铁路技术体系的研究[J].铁道运输与经济,2006(12):1-10.
[11] CANNY J.A computational approach to edge detection[J].IEEE Transactions on Pattern Analysis & Machine Intelligence,1986,8(6):679-698.
[12] STARK J A.Adaptive image contrast enhancement using generalizations of histogram equalization[J].IEEE Transactions on Image Processing,2000,9(5):889-896.
[13] 曹爽.基于Otsu算法的图像阈值分割技术[D].太原:太原理工大学,2018.
[14] 刘春阁.基于Hough变换的直线提取与匹配[D].阜新:辽宁工程技术大学,2009.
[15] 段汝娇,赵伟,黄松岭,等.一种基于改进Hough变换的直线快速检测算法[J].仪器仪表学报,2010,31(12):2774-2780.
[16] 罗文婷,李中轶,李林,等.基于改进Canny算法的道路标线自动识别及定位[J].西南交通大学学报,2018(12):1253-1260.
[17] 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.
[2] 陈金胜,陈兴杰,彭乐乐,等.基于边缘特征的无砟轨道扣件定位方法[J].计算机测量与控制,2018,26(5):172-175.
[3] LI X,ZHANG H.An improved canny edge detection algorithm[C]//2017 8th IEEE International Conference on Software Engineering and Service Science.Beijing:IEEE Press,2017:24-26.
[4] 周咏,顾桂梅,冉建民.基于改进主动轮廓模型的钢轨表面缺陷分割[J].兰州交通大学学报,2018,37(3):55-63.
[5] XIA Y Q,XIE F Y,JIANG Z Q.Broken railway fastener detection based on adaboost algorithm[C]//International Conference on Optoelectronics and Image Processing.New York:IEEE Press,2010:313-316.
[6] 张洋.巡道车视频处理系统关键技术的研究[D].北京:北京交通大学,2008.
[7] 范宏.基于图像的铁路扣件缺陷检测算法研究[D].成都:西南交通大学,2012.
[8] 万超.轨道扣件图像定位和特征识别算法研究[D].南昌:南昌大学,2015.
[9] 吴禄慎,万超,陈华伟,等.一种改进的十字交叉轨道扣件定位方法[J].铁道标准设计,2016(12):49-53.
[10] 何华武.建立中国高速铁路技术体系的研究[J].铁道运输与经济,2006(12):1-10.
[11] CANNY J.A computational approach to edge detection[J].IEEE Transactions on Pattern Analysis & Machine Intelligence,1986,8(6):679-698.
[12] STARK J A.Adaptive image contrast enhancement using generalizations of histogram equalization[J].IEEE Transactions on Image Processing,2000,9(5):889-896.
[13] 曹爽.基于Otsu算法的图像阈值分割技术[D].太原:太原理工大学,2018.
[14] 刘春阁.基于Hough变换的直线提取与匹配[D].阜新:辽宁工程技术大学,2009.
[15] 段汝娇,赵伟,黄松岭,等.一种基于改进Hough变换的直线快速检测算法[J].仪器仪表学报,2010,31(12):2774-2780.
[16] 罗文婷,李中轶,李林,等.基于改进Canny算法的道路标线自动识别及定位[J].西南交通大学学报,2018(12):1253-1260.
[17] 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.