Quickly and conveniently identifying buildings in disaster areas plays an important role in disaster assessment. To achieve the technical requirements of flood disaster relief projects, this paper proposes a building extraction method for use with remote sensing images that combines traditional digital image processing methods and convolution neural networks. First, the threshold segmentation method is used to select and construct a training dataset. Then, a variety of preprocessing methods are used to enhance the selected dataset. Finally, the improved Mask R-CNN algorithm is used to detect buildings in the images. Experiments show that compared to the R-CNN algorithm, the proposed method improves detection accuracy and reduces the computational time.

快速方便地识别灾区建筑物在灾难评估中起着重要作用。为了达到洪水救灾工程的技术要求，本文提出了一种结合传统数字图像处理方法和卷积神经网络的遥感图像建筑物提取方法。首先，阈值分割方法用于选择和构建训练数据集。然后，可以使用多种预处理方法来增强所选数据集。最后，使用改进的Mask R-CNN算法检测图像中的建筑物。实验表明，与R-CNN算法相比，该方法提高了检测精度，减少了计算时间。