联合改进U-Net模型和D-InSAR技术采矿沉陷提取方法

林佳惠, 刘广, 范景辉, 赵红丽, 白世彪, 潘宏宇

Extracting information about mining subsidence by combining an improved U-Net model and D-InSAR

LIN Jiahui, LIU Guang, FAN Jinghui, ZHAO Hongli, BAI Shibiao, PAN Hongyu

表3 不同模型在测试集上的精度评价

Tab.3 Accuracy evaluation of different models on test set

模型 PA/% MIoU /% IoU (背景) /% IoU (采矿 沉陷)/% 训练 时间/h FCN-32s 97.71 73.68 98.21 49.15 1.84 FCN-16s 98.16 76.61 98.38 54.84 1.88 FCN-8s 98.29 78.19 98.51 57.87 2.11 PSPNet 98.16 79.21 98.65 59.77 6.50 Deeplabv3 98.20 79.71 98.57 60.85 7.50 U-Net 98.31 79.24 98.27 60.20 5.08 本文模型 98.55 80.58 98.41 62.74 6.36