针对小目标要素的目视航空图增强方法

Enhancement Method of Visual Aeronautical Chart for Small Target Elements

增强地图是增强现实与地图学结合的新产物,由于跟踪注册算法不易识别小目标特征,已有研究多集中在全局的面状要素增强表达上。针对小目标点状符号的识别和增强设计两个问题,基于YOLOv3目标检测模型利用航空图数字影像数据实现以检索、阅读和多元化符号表达为主的目视航空图要素增强。通过深度卷积神经网络解决航空要素的符号识别问题;基于检测结果设计3类增强功能。经实验验证,该方法可以实时准确地增强航空要素,检测模型有较好的准确率和召回率,能满足目视航空图要素的小目标增强需求。

Augmented map is a new product of the combination of augmented reality and cartography. Due to the difficulty of tracking registration algorithm in recognizing small target features, most of the existing researches focus on the enhanced representation of global surface elements. In this paper, aiming at the recognition and enhancement design of small target point symbols, a visual aeronautical chart enhancement method based on YOLOv3 target detection model is implemented, which mainly focuses on retrieval, reading and diversified symbol expression. The problem of symbol recognition of aeronautical elements is solved by using deep convolutional neural network. Three kinds of enhanced functions are designed based on the detection results. Experimental results show that this method can enhance the aeronautical elements in real time, and the detection model has good accuracy and recall rate, which can meet the small target enhancement requirements of visual aeronautical map elements.