A new and improved method for road extraction from remote sensing images by fusing different scale features

doi:10.13474/j.cnki.11-2246.2022.0264

Abstract

Abstract: Aiming at the problem of inconspicuous fine road texture features and difficult information extraction in high-resolution remote sensing imagery, a new method of deep learning road extraction fusing different scale features is proposed and implemented. The method firstly introduces a CoT module to build a residual network to extract road features at different scales by making full use of local and global contextual information. Secondly, a feature pyramid attention module is built to fuse different levels of road feature information. Finally, a global attention upsampling module is used to recover road details in conjunction with the global context. The experimental results show that the proposed method is better than the existing methods in terms of recall and intersection ratio, and can extract the road information in remote sensing images more completely and accurately, which improves the efficiency of road extraction.

Key words: remote sensing image, road extraction, CoT module, feature pyramid, global attention

CLC Number:

P237

YAN Zhiheng, REN Chao, LI Yi, XU Ninghui, ZHANG Shengguo. A new and improved method for road extraction from remote sensing images by fusing different scale features[J]. Bulletin of Surveying and Mapping, 2022, 0(9): 58-62.

References

[1] 史文中, 朱长青, 王昱. 从遥感影像提取道路特征的方法综述与展望[J]. 测绘学报, 2001, 30(3):257-262.
[2] 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.
[3] 孙晨阳, 周廷刚, 陈圣波, 等. 基于矩形模板匹配的线状地物半自动提取方法研究[J]. 西南大学学报(自然科学版), 2015, 37(7):155-160.
[4] 连仁包, 王卫星, 李娟. 自适应圆形模板及显著图的高分辨遥感图像道路提取[J]. 测绘学报, 2018, 47(7):950-958.
[5] SCHUBERT H, VAN DE GRONDE J J, ROERDINK J B T M. Efficient computation of greyscale path openings[J]. Mathematical Morphology-Theory and Applications, 2016, 1(1):189-202.
[6] 王龙飞, 严春满. 道路场景语义分割综述[J]. 激光与光电子学进展, 2021, 58(12):1200002.
[7] ALSHEHHI R, MARPU P R. Hierarchical graph-based segmenta-tion for extracting road networks from high-resolution satellite images[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2017, 126:245-260.
[8] XU Yongyang, XIE Zhong, FENG Yaxing, et al. Road extraction from high-resolution remote sensing imagery using deep learning[J]. Remote Sensing, 2018, 10(9):1461.
[9] WANG Shuai, YANG Hui, WU Qiangqiang, et al. An improved method for road extraction from high-resolution remote-sensing images that enhances boundary information[J]. Sensors, 2020, 20(7):2064.
[10] 苏健民, 杨岚心, 景维鹏. 基于U-Net的高分辨率遥感图像语义分割方法[J]. 计算机工程与应用, 2019, 55(7):207-213.
[11] 吴仁哲, 蔡嘉伦, 刘国祥, 等. 针对高分影像的RDU-Net乡村路网提取方法[J]. 遥感信息, 2021, 36(1):29-36.
[12] 李代栋, 赫晓慧, 李盼乐, 等. 基于SPUD-ResNet的遥感影像道路提取网络[J]. 计算机工程与应用, 2021, 57(23):240-247.
[13] 贺浩, 王仕成, 杨东方, 等. 基于Encoder-Decoder网络的遥感影像道路提取方法[J]. 测绘学报, 2019, 48(3):330-338.
[14] XIE Saining, GIRSHICK R, DOLLÁR P, et al. Aggregated residual transformations for deep neural networks[C]//Proceedings of 2017 IEEE Conference on Computer Vision and Pattern Recognition. Honolulu, HI:IEEE,2017.

[1]	WU Hongrong, ZHU Lanwei, SHI Dong. Object-oriented extraction of Sanya coral reef sediment information [J]. Bulletin of Surveying and Mapping, 2022, 0(9): 63-67.
[2]	FENG Weiming, ZHANG Xinchang, SUN Ying, JIANG Ming, GAN Qiao, HOU Xingxing. High-resolution remote sensing image change detection network with Transformer structure [J]. Bulletin of Surveying and Mapping, 2022, 0(8): 36-40,92.
[3]	HUANG Yuhong, ZHOU Weixun. Similarity method for high-resolution remote sensing scene change detection [J]. Bulletin of Surveying and Mapping, 2022, 0(8): 48-53.
[4]	LIU Xiaozhong, FENG Cunjun, DENG Xiaoyuan, ZHAN Yuanzeng, YU Hui, WANG Xingkun. Building height extraction method based on sub-meter remote sensing image and its verification [J]. Bulletin of Surveying and Mapping, 2022, 0(8): 128-132.
[5]	LI Maolin, YAN Qingwu, ZHONG Xiaoya. A compound method for correcting vegetation information [J]. Bulletin of Surveying and Mapping, 2022, 0(6): 49-54142.
[6]	REN Yuejuan, GE Xiaosan. An road synthesis extraction method of remote sensing image based on improved DeepLabV3+ network [J]. Bulletin of Surveying and Mapping, 2022, 0(6): 55-61.
[7]	ZHANG Gongquan, YUE Jianping, LIU Shengnan, WANG Hailong, YANG Zhixiang, ZHOU Hangyu. Terrain extraction method of Poyang Lake based on remote sensing image [J]. Bulletin of Surveying and Mapping, 2022, 0(5): 62-66,73.
[8]	PAN Jianping, FU Zhanbao, CHENG Weihua, LU Shian, LI Xin, SUN Bowen, HUANG Guiping. Evaluation of mountain sponge city ecological construction based on landcover types [J]. Bulletin of Surveying and Mapping, 2022, 0(5): 84-88.
[9]	CHEN Zhilang, FU Zhenhua, ZHU Ziyang, WANG Huihui, LIU Qinwen, YANG Yuling, XU Gengran. HRNet-based extraction of building change information from high-resolution remote sensing images [J]. Bulletin of Surveying and Mapping, 2022, 0(5): 126-132.
[10]	LIU Xiaoyu, LIU Yang, DU Mingyi, ZHANG Min, JIA Jingjue, YANG Heng. Research on construction and demolition waste stacking point identification based on DeeplabV3+ [J]. Bulletin of Surveying and Mapping, 2022, 0(4): 16-19,43.
[11]	WANG Xue, LIANG Ke, SUI Lichun, ZHONG Mianqing, ZHU Jianfeng. Rural buildings extraction based on deep learning model with dilated convolution and pyramid representation [J]. Bulletin of Surveying and Mapping, 2022, 0(4): 61-65.
[12]	GAO Ming, ZHOU Xinxin, LIU Qi, YANG Guangdi, WU Changbin. Automatic monitoring of farmland occupation by farm house based on deep learning network [J]. Bulletin of Surveying and Mapping, 2022, 0(3): 47-53.
[13]	WANG Shengli, ZHU Shouhong, JIANG Yi. Scene recognition of photovoltaic panels based on model migration and convolution neural network [J]. Bulletin of Surveying and Mapping, 2022, 0(2): 5-9.
[14]	YANG Sufang, DU Lin. High-resolution remote sensing image classification based on multi-feature popular discrimination embedding [J]. Bulletin of Surveying and Mapping, 2021, 0(9): 37-42.
[15]	GENG Wanxuan, ZHOU Weixun, JIN Shuanggen. Scene classification of high-resolution remote sensing image based on multi-kernel SVM using features extracted from LDCNN [J]. Bulletin of Surveying and Mapping, 2021, 0(8): 14-21.