Fig. 1. Flow chart of building contour regularization

Fig. 2. Schematic of building flat rotation transformation algorithm based on MBR. (a) Flow chart; (b) process diagram

Fig. 3. Process of corners acquisition based on Shi-Tomasi. (a) Angle relationship of corners; (b) corners acquisition result; (c) corners filtering result

Fig. 4. Schematic of contour reconstruction by local optimal weight fitting. (a)(d) Segmentation points acquisition (red pentagram) based on Shi-Tomasi and MBR; (b) local segment recombination; (e) MBR; (c)(f) regularized contour

Fig. 5. Schematic of attribute allocation and reorganization for local segment. (a) Attribute assignment; (b) merging constraint for the same attribute; (c) minimum distance threshold constraint

Fig. 6. High-resolution remote sensing image. (a) Image S; (b) extraction result; (c) reference data

Fig. 7. Accuracy and efficiency indexes corresponding to different ratio thresholds. (a)(b) Accuracy evaluation; (c) efficiency analysis; (d) number of rectangular categories

Fig. 8. Schematic of contour regularization of different methods. (a) Method A; (b) method B; (c) method C; (d) method D; (e) proposed method

Fig. 9. Comparison of single building contour regularization effect. (a) Label data; (b) extraction result; (c) method A; (d) method B; (e) method C; (f) method D; (g) proposed method

Fig. 10. Regularization effect of irregular building contour. (a) Original images; (b) SVM classification extraction results; (c) regularization results; (d) overlay of regularization results and original images