基于ORB算法的输电线路异物识别研究

针对输电线路异物检测问题,通过分析航拍视频,提取关键帧并采用帧差法标注异物,使用特征点跟踪异物,从而达到输电线路异物检测的目的。将预估区域漂移法与欧式距离法融合,弥补关键帧冗余的缺陷;使用概率密度函数分析网格内异物占有率,剔除帧差法中微小非目标区域;提出K-means算法聚类分析Oriented Brief(ORB)算子,可精简特征点提高匹配率。实验结果表明,能有效精简关键帧,并且改进帧差法可精确提取异物;同时快速准确提取ORB算子;故而可快速识别线上异物。

The Recognition and Tracking of Foreign Body on Transmission Line Based on ORB Algorithm

The research of this paper solve the question of foreign body on transmission line.It put forward a algorithm to extract key frames and using frame difference method to indicate foreign body,then tracking foreign body based on features point by analyzing aerial video.Fusing euclidean distance method and forecasting the drift of area,which can remedy the defect of key frames redundancy.Via analyzing probability density function to eliminate the small target area.Eventually proposing K-means algorithm to cluster analysis the ORiented Brief(ORB) features which can improve the matching accuracy. Experimental results show that this method can streamline key frames. The foreign body is accurately identified by improving frame difference method. And the speed of extracting ORB features is lifted in this paper. In conclusion, foreign body online is quickly and accurately identified by the method in this paper.