中文摘要：

针对货运列车缓解阀盖缺失故障的检测,提出一种基于轮廓有向线段重构与成对几何直方图的匹配方法。该方法首先根据最大弦长和极半径确定采样初始位置,并根据局部弯曲度评价机制对轮廓进行动态分级采样;然后以采样点为端点按逆时针顺序构建有向线段,依次计算每对线段间的有向相对角和归一化当量距离,并最终将其作为该对线段的双重特征描述子计入二维直方图;最后使用巴氏距离对二维直方图间的相似性进行度量。实验分析表明,该算法对旋转、缩放和平移等几何变换有较好的鲁棒性,同时也兼顾了检测效率,满足了列车故障检测的实时性要求。

英文摘要：

For the missing faults of release valve bonnets in freight cars,an image matching algorithm was proposed herein based on contour reconstructed by oriented line segment and pairwise geometrical histogram.Firstly,the location of initial sampling point was defined by maximum chord length and polar radius,then the contour was dynamically and hierarchically sampled according to an evaluation mechanism of local curvatures.Then adjacent sample points were lined in counter-clockwise order,oriented relative angles and normalized equivalent distances between pairwise oriented line segment were calculated as the dual descriptors,which were counted into the two dimensional histogram.Finally the similarity between histograms was measured by Bhattacharyya distance.The experimental results show that the algorithm may keep robust under the circumstances of rotations,scales and translations,meanwhile the detection efficiency is also well ensured,which may satisfy the instantaneity of the detection on freight cars.