基于AlexNet网络的高分辨率遥感影像建筑物提取方法研究

Research on Building Extraction Method from High Resolution Remote Sensing Images Based on AlexNet Network

高分辨率遥感影像的数据源日益增长使得其成为主要的遥感数据源之一。本文研究了一种基于AlexNet网络的高分辨率遥感影像建筑物提取方法,该方法是在卷积神经网络的基础上,建立一种端到端自动提取影像中建筑等物体位置的方法:首先使用图像增广技术增加数据集的丰富性和多样性;再通过超参数搜索选择网络使用的较优参数,最终实现了遥感影像中建筑物的自动提取。实验结果表明,该方法可达到75%的提取精度;与传统方法进行定性和定量对比,该方法具有耗时少、精度高的特性,对后续城市规划、三维建模等应用有着重要意义。

The growing number of data sources of the high resolution remote sensing images makes it one of the main remote sensing data sources.This paper studies a kind of method that extracts the building from high resolution remote sensing images using the AlexNet network.Based on the convolution neural network,this thesis establishes an end-to-end method that automatically extracts the position of the buildings as well as other objects from images:Firstly,the image enhancement techniques are used to increase the richness and diversity of the dataset;Then the hyperparameter search is implemented to decide the optimal parameters used in the network,and finally realizes the automatic extraction of buildings in remote sensing image.The results show that the proposed method can achieve the extraction accuracy of 75%,and the qualitative and quantitative comparison with the traditional method shows that the proposed method has the characteristics of less time consuming and high precision,which is of great significance to the subsequent application of urban planning and 3D modeling.