Full convolutional network building extraction method based on contour guidance and structural attention

Abstract

The invention discloses a full convolutional network building extraction method based on contour guidance and structural attention. The method comprises the following steps: acquiring a remote sensing image; inputting the remote sensing image into a trained full convolutional network based on contour guidance and structural attention, and outputting a building strength graph through the full convolutional network based on contour guidance and structural attention; and based on a threshold algorithm, extracting a binary image of the building strength graph to obtain a building extraction result. According to the full convolutional network based on contour guidance and structural attention, the edge information of the building can be better concerned through contour guidance and structural attention, so that the edge extraction precision of the building can be improved, and the full convolutional network has relatively good robustness and practicability. 本发明公开了基于轮廓引导和结构注意的全卷积网络建筑物提取方法，所述方法包括：获取遥感图像；将所述遥感图像输入已训练的基于轮廓引导和结构注意的全卷积网络，通过基于轮廓引导和结构注意的全卷积网络输出建筑物强度图；基于阈值算法，提取所述建筑物强度图的二值图像，得到建筑物提取结果。本发明实施例的基于轮廓引导和结构注意的全卷积网络通过轮廓引导和结构注意可以更好的关注建筑物的边缘信息，从而能够提高建筑物提取边缘精度，并且具有较好的鲁棒性和实用性。