基于时空上下文的航拍视频绝缘子跟踪算法

Aerial Video Insulator Tracking Algorithm Based on Spatio-temporal Context

为解决时空上下文跟踪算法在航拍视频绝缘子出现大尺度变化产生的漂移问题,提出一种大尺度自适应的航拍视频绝缘子跟踪算法。提取绝缘子图像的局部二值模式特征构建时空上下文模型,改进时空上下文尺度更新策略不能适应大尺度变化问题。最后在航拍视频序列进行的对比实验表明,算法在平均重叠率和平均中心位置误差指标上有了较大提升;且能很好地适应绝缘子大尺度变化,具有更好的跟踪效果。

In order to solve the drift problem of spatio-temporal context tracking algorithm in large-scale variation of aerial video insulators,a large-scale adaptive tracking algorithm for aerial video insulators was proposed. Local binary pattern feature of insulator images was extracted to construct spatio-temporal context model,and improved spatio-temporal context updating strategy cannot adapt to large scale changes. Finally,the comparative experiments on aerial video sequences show that the proposed algorithm has greatly improved the average overlap rate and the average central location error,and also can well adapt to large-scale insulator variation,with better tracking effect.