基于关键特征梯形的城市轨道交通车辆受电弓羊角形变检测方法

Detection Method of Urban Rail Transit Vehicle Pantograph Contact Strip Deformation Based on Key Feature Trapezoid

针对城市轨道交通车辆受电弓羊角的形变和断裂,提出了基于关键特征梯形的受电弓羊角形变检测方法。首先,进行图像滤波预处理,去除图片中的高斯噪声;其次,对预处理图像进行LC(亮度对标)显著性检测,将图像的前景和背景分离,利用GrabCut算法完成图像分割,简化除采集图像中的受电弓以外的复杂环境背景,保留受电弓的完整特征;最后,按一阶梯度和二阶梯度信息为特征关键点匹配模板,得到2个羊角尖和2个螺母的关键点位置,将4个关键点连接形成特征梯形,对特征梯形面积进行计算,并与标准的受电弓特征面积进行比对,进而计算出受电弓羊角的形变比。试验结果表明:该算法能准确地定位羊角位置和计算其形变,可为维修人员及时更换受电弓提供依据。

Targeting the deformation/breakage of urban rail transit pantograph contact strip,a deformation detection method of it based on key feature trapezoid is proposed.First,image filtering preprocessing is performed to remove Gaussian noise in the image.Then,LC(luminance coefficient)saliency detection is performed on the preprocessed image to separate the foreground and background from the image,and GrabCut algorithm is used to complete the image segmentation,simplifying the complex environment around pantograph in the image to preserve the complete features of it.Finally,the feature key point template is matched according to the first-order gradient and second-order gradient information,and the key point positions of 2 contact strip ends and 2 nuts are obtained.The four key points are connected to form a feature trapezoid,its area is calculated and compared with the standard pantograph feature area,the deformation ratio of the pantograph contact strip end can be further obtained.The test results show that the algorithm can accurately locate the contact strip end and calculate its deformation,providing a basis for maintenance personnel to replace the pantograph in time.