摘要
随着无人飞行器的大范围普及,针对其场景理解的技术和应用也变得越来越重要。论文采用BOO原理建立语义标注模型,利用地图软件中显示的瓦片金字塔技术,结合事件识别领域提供的训练数据集和测试数据集,提出了一种新的语义分层方法。实验仿真证明该方法能够使无人机在不同高度拍摄的场景中识别出相应的景物。
With the widespread popularity of UAV,UAV scene understanding technology is more and more important. In thispaper,BOO principle is used to establish semantic annotation model,and the tile-pyramid technology of map software is used andthe training dataset and test dataset provided in event recognition field are combined,and a new semantic layering method is pro-posed. The experimental simulation proves that the proposed method can identify the corresponding objects in the scene of the UAVphotographed at different heights.
出处
《舰船电子工程》
2017年第10期83-86,共4页
Ship Electronic Engineering
关键词
场景理解
视觉字典
语义标注
语义分层
瓦片金字塔
scene understanding
texton map
semantic annotation
semantic layering
tile-pyramid