基于SURF特征匹配的电气化铁路接触网支撑装置旋转双耳不良状态检测

Defective Condition Detection of Rotary Double Ears of Support Device in Electrified Railway Based on Feature Matching of SURF

针对电气化铁路接触网支撑装置中旋转双耳耳片断裂问题，提出一种基于快速鲁棒性特征（Speeded—Up Robust Features，SURF）匹配的图像检测方法。首先利用标准旋转双耳图像与待检测图像局部不变特征点的匹配，实现旋转双耳的初识别；然后应用Hough变换实现耳片精确定位；最后通过耳片局部图像中的灰度方差判断耳片是否存在断裂状态。实验表明，该方法能在复杂图像中较准确地识别耳片断裂特征，为旋转双耳的状态检测提供一定参考。

A novel image detection method based on speeded -up robust features （SURF） was proposed, which aimed at detecting the break of rotary double ears of support device in electrified railway. Firstly, the matching invariant feature point between standard image and sampled image to be detected to realize the initial recognition of rotary ears. Secondly, the ears were accurately located through Hough transform. Lastly, the condition of ears was able to be distinguished by the gray variance of ears local image. The experiments show that this method can accurately identify the break features of ears in the intricate graphics, and can offer a reference to the state detection of rotary double ears.