Building Contour Optimization Method based on V-System

Building extraction technology is important in urban construction planning, map mapping, promoting remote sensing image mining technology development, etc. However, the current deep learning extraction of building boundary information in remote sensing images has problems such as jagged edges and irregular building polygons caused by the noise interference of surrounding objects. This paper proposes a buildings contour optimization method based on the v-system. Firstly, based on the results of obtaining the building boundary information using the deep complete convolutional network, the building is subjected to canny operator edge detection to obtain the edge point array information of the building. Then, the building outline is initially fitted by using V-system, and then the polygon feature point information is obtained by combining the clustering analysis algorithm to classify the fitting results. Finally, according to the extracted feature points, the optimization of the building boundary contour is realized, and the accuracy of the edge information is improved. The experimental results show that this method not only improves the boundary jaggedness problem but also realizes the extraction of overall features and local detail features of the building contour and further improves the consistency with the actual building contour.