Journal of Geo-information Science >
Road Extraction Model of High-resolution Remote Sensing Images based on Dual-attention Residual Network
Received date: 2022-07-14
Revised date: 2022-09-08
Online published: 2023-04-19
Supported by
National Natural Science Foundation of China(41971414)
In high-resolution remote sensing images, the spectral information of road is rich, and the spatial geometric structure is clear. However, the road extraction is still faced with challenges such as changes in road size and influences from trees, buildings, and occlusion shadow, which often leads to incomplete extraction results. In addition, the phenomenon of the same object with different spectrum and the foreign body with the same spectrum is more serious, which affects the continuity of road extraction and the integrity of small road information, and it is difficult to distinguish road and non-road impervious layer. Therefore, a road extraction model, DARNet, is proposed in this study to address the above limitations. It uses a deep learning network to obtain fine-grained high-level semantic information and enhance the network's ability to extract fine roads. By embedding the serial channel-space dual attention module, the global semantic information of road feature map is obtained, and the robustness of the network model in shadow and occlusion environment is enhanced. The efficient expression of road features and the deep fusion of multi-scale road information are achieved, the phenomenon of missing details in road extraction is improved, and the efficient and accurate automatic road extraction in complex environment is realized. In this paper, a quantitative comparison is carried out based on three experimental datasets, using DARNet, DLinkNet, and DeepLabV3+ etc. The results show that the F1 of the proposed model is 77.92%, 67.88% and 80.37% for three datasets, respectively, which is higher than that of the comparison models. In addition, the qualitative comparison shows that the proposed model can effectively overcome the problem of inaccurate and incomplete road extraction caused by object shadow, occlusion, and spectral changes of high-resolution images, avoid the phenomenon of missing and miscarrying of small roads, and improve the integrity and continuity of road network extraction.
LIU Yang , KANG Jian , GUAN Haiyan , WANG Hanyun . Road Extraction Model of High-resolution Remote Sensing Images based on Dual-attention Residual Network[J]. Journal of Geo-information Science, 2023 , 25(2) : 396 -408 . DOI: 10.12082/dqxxkx.2023.220513
表1 实验数据集参数统计Tab. 1 Statistical table of experimental data sets parameters |
数据集 | 年份 | 影像大小/像素 | 空间分辨率/m | 影像数量/张 |
---|---|---|---|---|
Massachusetts | 2013 | 1500×1500 | 1 | 1171 |
DeepGlobe | 2018 | 1024×1024 | 0.5 | 6226 |
CHN6-CUG | 2021 | 512×512 | 0.5 | 4511 |
表2 DARNet和对比模型在实验数据集上提取结果统计Tab. 2 DARNet and the comparison model extract the result tables on the experimental dataset (%) |
方法 | Massachusetts | DeepGlobe | CHN6-CUG | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Pre | F1 | mIoU | Pre | F1 | mIoU | Pre | F1 | mIoU | |||
DARNet | 78.81 | 77.92 | 80.43 | 98.56 | 67.88 | 73.91 | 81.82 | 80.37 | 81.25 | ||
DLinkNet | 77.25 | 74.48 | 78.22 | 97.16 | 64.66 | 71.39 | 80.48 | 77.43 | 79.02 | ||
DeepLabv3+ | 77.89 | 72.79 | 77.09 | 97.07 | 64.42 | 71.24 | 76.82 | 72.38 | 75.30 | ||
FCN8s | 72.30 | 68.80 | 57.96 | 96.14 | 65.17 | 71.68 | 73.97 | 70.32 | 73.81 | ||
SegNet | 78.52 | 73.25 | 57.79 | 96.85 | 61.66 | 69.63 | 76.99 | 71.96 | 75.02 | ||
UNet | 78.78 | 74.50 | 78.24 | 96.94 | 67.09 | 72.90 | 77.57 | 74.72 | 76.95 |
图6 DARNet和对比模型在Massachusetts数据集的道路提取结果Fig. 6 DARNet and Contrast model for road extraction results on Massachusetts dataset |
图7 DARNet和对比模型在DeepGlobe数据集的道路提取结果Fig. 7 DARNet and Contrast model for road extraction results on DeepGlobe dataset |
表3 DARNet和消融模型在实验数据集提取结果统计 |
Massachusetts | DeepGlobe | CHN6-CUG | |||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Method | Pre | F1 | mIoU | Pre | F1 | mIoU | Pre | F1 | mIoU | ||
DARNet | 78.81 | 77.92 | 80.43 | 98.56 | 67.88 | 73.91 | 81.82 | 80.37 | 81.25 | ||
DR-DLinkNet | 77.42 | 74.68 | 78.36 | 97.33 | 65.03 | 71.54 | 80.57 | 77.70 | 79.22 | ||
DA-DLinkNet | 77.89 | 74.66 | 78.57 | 96.93 | 65.32 | 71.78 | 80.82 | 77.58 | 79.14 | ||
DLinkNet | 77.25 | 74.48 | 78.22 | 97.16 | 64.66 | 71.39 | 80.48 | 77.43 | 79.02 |
Tab. 3 DARNet and ablation model extracted results from experimental datasets in statistical tables (%) |
[1] |
|
[2] |
张永宏, 何静, 阚希, 等. 遥感图像道路提取方法综述[J]. 计算机工程与应用, 2018, 54(13):1-10,51.
[
|
[3] |
曹敏. 基于频谱的高分辨率遥感影像纹理尺度分析及选择[D]. 北京: 中国地质大学(北京), 2020.
[
|
[4] |
|
[5] |
|
[6] |
|
[7] |
|
[8] |
|
[9] |
|
[10] |
韩洁, 郭擎, 李安. 结合非监督分类和几何—纹理—光谱特征的高分影像道路提取[J]. 中国图象图形学报, 2017, 22(12):1788-1797.
[
|
[11] |
曹云刚, 王志盼, 杨磊. 高分辨率遥感影像道路提取方法研究进展[J]. 遥感技术与应用, 2017, 32(1):20-26.
[
|
[12] |
|
[13] |
|
[14] |
|
[15] |
戴激光, 王杨, 杜阳, 等. 光学遥感影像道路提取的方法综述[J]. 遥感学报, 2020, 24(7):804-823.
[
|
[16] |
|
[17] |
|
[18] |
|
[19] |
|
[20] |
|
[21] |
|
[22] |
|
[23] |
|
[24] |
|
[25] |
|
[26] |
|
[27] |
|
[28] |
|
[29] |
|
[30] |
|
[31] |
宋廷强, 刘童心, 宗达, 等. 改进U-Net网络的遥感影像道路提取方法研究[J]. 计算机工程与应用, 2021, 57(14):209-216.
[
|
[32] |
|
[33] |
刘航, 汪西莉. 基于注意力机制的遥感图像分割模型[J]. 激光与光电子学进展, 2020, 57(4):11.
[
|
[34] |
|
[35] |
|
[36] |
|
[37] |
|
[38] |
|
/
〈 |
|
〉 |