A Saliency Detection Method for Extracting Buildings from High-Resolution Remote Sensing Images
Aiming at the characteristics of buildings on high-resolution remote sensing images,a deep learning saliency detection method for extracting buildings on high-resolution remote sensing images is designed and implemented.Firstly,FCN semantics is used to segment high-resolution remote sensing images,and the difference between the mid-level features of each pixel is calculated according to the mid-level features of the high-resolution remote sensing image,such as uniqueness,compactness and background,and the feature map of each pixel is obtained,and these feature maps are combined together to obtain the global prior feature map;Secondly,the saliency detection target method is proposed,that is,the global prior significance map is obtained after the global prior feature map is input into the deep learning network model;Finally,through experimental data,the accuracy and recall rate P-R curve and F-Measure are used as evaluation indicators to verify the method and analyze the results.It is shown that the proposed method can completely and accurately extract high-resolution remote sensing images buildings with fast calculation speed,excellent performance,and the process of extracting element features without manual involvement,and have good application value.
deep learningsaliency detectionremote sensing imagebuilding extraction
NETL
NSTL
万方数据
