中文摘要：

针对列车集尘器定位不准确的问题，提出了一种基于几何特征的形状匹配算法。该算法首先对轮廓点进行采样，基于极半径、局部曲率确定关键点的初始位置及点集的映射关系，然后以形心为基准，生成以角度和尺度为几何特征的双重描述子，并对其作标准量化处理。最后使用改进的曼哈顿距离计算描述子的相似性。实验结果表明，该形状匹配算法几乎不受伸缩、旋转、平移等几何变换的影响，具有一定的适应性和鲁棒性。

英文摘要：

Due to the problem of inaccurate positioning of the train dust collector, a shape matching method was proposed based on geometric features. Firstly, the contour points were sampled and initial location of critical points and mapping relations of point sets were determined based on polar radius and local curvature. Then taking centroid as a benchmark,dual descriptors with geometric features of angles and scales were generated and processed by standard quantification. Finally, the similarity of the descriptors was calculated with modified Manhattan distance. The experimental results show that the proposed shape matching method is harldy influenced by geometric transformation such as scale, rotation,and translation, which demonstrates its adaptability and robust.