基于特征匹配的接触网定位支座检测

Detection of Catenary Positioning Bearing Based on Feature Matching

针对高速铁路接触网的定位支座识别问题,提出利用图像处理技术中的特征匹配的方法对接触网综合检测车采集的接触网图像进行定位支座的检测,采用3种常见的特征描述子分别对定位支座进行旋转不变和尺度不变的特征点检测,对SURF和ORB特征描述点用欧式距离进行匹配,采用汉明距离对FREAK描述子进行匹配。试验表明SURF的匹配效果最好,性能最佳,该方法能够进行实际运用。

Aiming at the problem of positioning bearing identification of high-speed railway catenary, using feature matching method in image processing technology to locate the catenary bearing by catenary comprehensive detection vehicle was proposed, and three kinds of characteristic descriptors were used to detect the feature points with constant rotation and constant scale. SURF and ORB were matched with Euclidean distance, and the FREAK descriptor was matched by Hamming distance. Experiments showed that SURF has the best matching effect and the best performance. The method can be used in practice.