针对建筑轮廓提取问题，提出并实现了一种基于MATLAB的影像建筑轮廓自动提取方法。首先，对彩色影像进行阴影去除，将彩色影像转换为灰度影像，进行直方图均匀化以及中值滤波；其次，对中值滤波后的影像进行膨胀和腐蚀操作，通过对膨胀与腐蚀影像进行差分运算，进行地物边缘检测；最后，对边缘检测后的影像二值化处理，并完成建筑物特征点及其坐标的提取。通过实例验证表明，本文方法简洁可行，计算效率要高于深度学习提取建筑物的方法。
Aiming at the problem of building contour extraction, an automatic extraction method of building contour based on MATLAB is proposed and implemented. Firstly, the shadow of the color image is removed, the color image is converted into gray image, histogram homogenization and median filtering are carried out; secondly, the image after median filtering is expanded and corroded, and the ground edge is detected by differential operation of the expansion and corrosion image; finally, the image after edge detection is binarized, and then the building feature points and their coordinates are extracted. The example shows that this method is simple and feasible, and the computational efficiency is much higher than the deep learning method.
Huang, X. and Zhang, L. (2013) An SVM Ensemble Approach Combining Spectral, Structural, and Seman-tic Features for the Classification of High-Resolution Remotely Sensed Imagery. IEEE Transactions on Geoscience and Remote Sensing, 51, 257-272. https://doi.org/10.1109/TGRS.2012.2202912
