All Issue

2025 Vol.60, Issue 2 Preview Page

Research Article

AI-Based Road Extraction Considering Road Structural Characteristics

구조적 특성을 고려한 AI 기반 도로 추출

Jaegwan Leea
,
Jinmu Choib*
이 재관a
,
최 진무b*
aMaster Course, Department of Geography, Kyung Hee University
bProfessor, Department of Geography, Kyung Hee University
a경희대학교 지리학과 석사과정
b경희대학교 지리학과 교수
*Corresponding Author
30 April 2025. pp. 187-196
PDF XML Citation
Abstract

Recent advancements have led to an increased necessity for rapid and accurate updates of road information, wherein remote sensing imagery and deep learning-based semantic segmentation techniques are playing a crucial role in road extraction research.However, when applying deep learning segmentation techniques, issues arise due to occlusions from elements such as trees and building shadows, resulting in incomplete learning of the road's structure and connectivity in the extraction results. Therefore, this study aims to propose a method for road extraction based on artificial intelligence by utilizing masking data in order to consider the structural characteristics of roads. To achieve this, the present study employed artificial masks at the edges of the road as the training data for deep learning, enabling the model to exclude the effects of being obscured by trees and shadows at the road's edge during the training process. To this end, the deep learning model was trained under three conditions regarding 15,000 training data instances: without the application of occlusion masks, with the generation of 50% occlusion masks, and with the generation of 100% occlusion masks, followed by a comparative analysis of the road extraction results. The deep learning models utilized in this study include U-Net, SegNet, and SegFormer. For the structural evaluation of the road extraction results, pixel-based evaluation metrics (Accuracy, Precision, Recall, F1- Score, and mIoU) were employed, alongside APLS (Average Path Length Similarity) and SSIM (Structural Similarity Index), which reflect the structural characteristics of the roads. The analysis results indicate that the SegFormer, utilizing 50% of the masking data, exhibited the most superior performance. Therefore, it has been confirmed that the structural characteristics of land cover can be maintained through deep learning models utilizing masking data. In the future, to minimize the issues arising from structural features being obscured by objects such as trees or shadows, it is anticipated that employing artificial masks during the training process of deep learning models will enhance the accuracy of feature extraction.

Keywords
Road Extraction
Remote Sensing
Semantic Segmentation
Structural Evaluation
Masking Data

최근 신속하고 정확한 도로 정보 갱신의 필요성이 증가함에 따라 원격탐사 영상과 딥러닝 기반 의미론적 분할 기법이 도로 추출 연구에서 중요한 역할을 하고 있다. 하지만 딥러닝 분할 기법을 적용할 때 나무, 건물의 그림자 등으로 인해 도로의 일부가 가려져 추출 결과에서 도로의 구조와 연결성이 온전하게 학습되지 못하는 문제를 겪는다. 따라서 본 연구는 도로의 구조적 특성을 고려하기 위해 마스킹 데이터를 활용하여 AI 기반의 도로 추출 방안을 제안하고자 하였다. 이를 위해 딥러닝의 훈련 데이터로 도로의 가장자리에 인위적인 가림 마스크를 넣어 훈련과정에서 모델이 도로의 가장자리에서 나무와 그림자로 가려지는 효과를 배제할 수 있도록 하였다. 이를 위해 15,783개의 훈련 데이터에 대해 가림 마스크를 적용하지 않은 경우, 가림 마스크를 적용한 데이터를 전체 데이터에서 50% 섞은 경우, 가림 마스크를 적용한 데이터로만 훈련 데이터셋을 구성한 경우로 구분하여 딥러닝 모델을 학습하고 도로 추출 결과를 비교하였다. 딥러닝 모델로는 U-Net, SegNet, SegFormer를 활용하였다. 도로 추출 결과의 구조적 평가를 위해 픽셀 기반의 평가 지표(Accuracy, Precision, Recall, F1-Score, and mIoU)와 함께 도로의 구조적 특징을 반영한 APLS(Average Path Length Similarity)와 SSIM(Stuctural Similarity Index)도 활용하였다. 분석 결과, 마스킹 데이터를 50% 활용한 SegFormer가 가장 우수한 성능을 보였다. 따라서 마스킹 데이터를 활용한 딥러닝 모델을 통해 토지피복의 구조적 특성을 유지할 수 있음을 확인하였다. 향후 도로와 같이 구조적 특성이 있는 사상이 나무나 그림자 등으로 가려지는 문제를 최소화하기 위해 딥러닝 모델 훈련 과정에서 인위적인 마스크를 활용하면 사상의 추출 정확도를 향상할 수 있을 것으로 기대한다.

키워드
도로추출
원격탐사
의미론적 분할
구조적 평가
마스킹 데이터
References
1

Babaali, K. O., Zigh, E., Djebbouri, M. and Chergui, O., 2022, A new approach for road extraction using data augmentation and semantic segmentation, Indonesian Journal of Electrical Engineering and Computer Science, 28(3), 1493-1501.

10.11591/ijeecs.v28.i3.pp1493-1501
2

Barzohar, M. and Cooper, D. B., 1996, Automatic finding of main roads in aerial images by using geometric- stochastic models and estimation, IEEE Transactions on Pattern Analysis and Machine Intelligence, 18(7), 707-721.

10.1109/34.506793
3

Chen, H., Hou, L., Zhang, G. K. and Wu, S., 2023, Using context-guided data augmentation, lightweight CNN, and proximity detection techniques to improve site safety monitoring under occlusion conditions, Safety Science, 158, 105958.

10.1016/j.ssci.2022.105958
4

Chen, P., Liu, S., Zhao, H., Wang, X., and Jia, J., 2020, Gridmask data augmentation, arXiv preprint arXiv: 2001.04086.

5

Chen, S. B., Ji, Y. X., Tang, J., Luo, B., Wang, W. Q. and Lv, K., 2021, DBRANet: road extraction by dual-branch encoder and regional attention decoder, IEEE Geoscience and Remote Sensing Letters, 19, 1-5.

10.1109/LGRS.2021.3074524
6

Chen, Z., Deng, L., Luo, Y., Li, D., Junior, J.M., Gonçalves, W. N., and Li, D., 2022, Road extraction in remote sensing data: a survey, International Journal of Applied Earth Observation and Geoinformation, 112, 102833.

10.1016/j.jag.2022.102833
7

DeVries, T., 2017, Improved regularization of convolutional neural networks with cutout, arXiv preprint arXiv: 1708.04552.

8

Feng, D., Shen, X., Xie, Y., Liu, Y. and Wang, J., 2021, Efficient occluded road extraction from high-resolution remote sensing imagery, Remote Sensing, 13(24), 4974.

10.3390/rs13244974
9

Guo, X. and Zhou, R., 2023, Data augmentation method for extracting partially occluded roads from high spatial resolution remote sensing images, IEEE Access, 11, 79232-79239.

10.1109/ACCESS.2023.3298550
10

Kass, M., Witkin, A. and Terzopoulos, D., 1988, Snakes: active contour models, International Journal of Computer Vision, 1(4), 321-331.

10.1007/BF00133570
11

Kim, B. R. and Oh, H., 2020, AI training dataset for cloud detection of KOMPSAT images, Geo Data, 2(2), 56-62.

10.22761/DJ2020.2.2.008
12

Li, R., Gao, B., and Xu, Q., 2020, Gated auxiliary edge detection task for road extraction with weight- balanced loss, IEEE Geoscience and Remote Sensing Letters, 18(5), 786-790.

10.1109/LGRS.2020.2985774
13

Lin, X., Shen, J., and Liang, Y., 2012, Semi-automatic road tracking using parallel angular texture signature, Intelligent Automation & Soft Computing, 18(8), 1009-1021.

10.1080/10798587.2008.10643306
14

Lin, X., Zhang, J., Liu, Z., and Shen, J., 2008. Semi-automatic extraction of ribbon roads from high resolution remotely sensed imagery by T-shaped template matching, In Geoinformatics 2008 and Joint Conference on GIS and Built Environment: Classification of Remote Sensing Images, 7147, 168-175.

10.1117/12.813220
15

Mei, J., Li, R. J., Gao, W., and Cheng, M. M., 2021, CoANet: connectivity attention network for road extraction from satellite imagery, IEEE Transactions on Image Processing, 30, 8540-8552.

10.1109/TIP.2021.311707634618672
16

Melgani, F. and Bruzzone, L., 2004, Classification of hyperspectral remote sensing images with support vector machines, IEEE Transactions on Geoscience and Remote Sensing, 42(8), 1778-1790.

10.1109/TGRS.2004.831865
17

Mena, J. B., 2003, State of the art on automatic road extraction for GIS update: a novel classification, Pattern Recognition Letters, 24(16), 3037-3058.

10.1016/S0167-8655(03)00164-8
18

Park, S. H., Lee, G. S., and Jung, H. S., 2017, Shadow detection and correction method for urban area using KOMPSAT-3 Image, Korean Journal of Remote Sensing, 33(6), 1197-1213.

19

Park, S. R., 2001, Semi-automatic road extraction algorithm from IKONOS images using template matching, In Proceedings of 22nd Asian Conference on Remote Sensing.

20

Senthilnath, J., Varia, N., Dokania, A., Anand, G., and Benediktsson, J. A., 2020, Deep TEC: deep transfer learning with ensemble classifier for road extraction from UAV imagery, Remote Sensing, 12(2), 245.

10.3390/rs12020245
21

Tan, Y. Q., Gao, S. H., Li, X. Y., Cheng, M. M., and Ren, B., 2020, Vecroad: point-based iterative graph exploration for road graphs extraction, In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 8910-8918.

10.1109/CVPR42600.2020.00893
22

Van Etten, A., Lindenbaum, D., and Bacastow, T. M., 2018, Spacenet: a remote sensing dataset and challenge series, arXiv preprint arXiv: 1807.01232.

23

Wang, Y., Seo, J., and Jeon, T., 2021, NL-LinkNet: toward lighter but more accurate road extraction with nonlocal operations, IEEE Geoscience and Remote Sensing Letters, 19, 1-5.

10.1109/LGRS.2021.3050477
24

Wang, Z., Bovik, A. C., Sheikh, H. R., and Simoncelli, E. P., 2004, Image quality assessment: from error visibility to structural similarity, IEEE Transactions on Image Processing, 13(4), 600-612.

10.1109/TIP.2003.81986115376593
25

Wei, Y., Zhang, K., and Ji, S., 2020, Simultaneous road surface and centerline extraction from large-scale remote sensing images using CNN-based segmentation and tracing, IEEE Transactions on Geoscience and Remote Sensing, 58(12), 8919-8931.

10.1109/TGRS.2020.2991733
26

Yager, N., and Sowmya, A., 2003, Support vector machines for road extraction from remotely sensed images, In Proceedings of International Conference on Computer Analysis of Images and Patterns, Berlin, Heidelberg: Springer Berlin Heidelberg, 285-292.

10.1007/978-3-540-45179-2_36
27

Yang, F., Wang, H., and Jin, Z., 2020, A fusion network for road detection via spatial propagation and spatial transformation, Pattern Recognition, 100, 107141.

10.1016/j.patcog.2019.107141
28

Yang, R., Zhong, Y., Liu, Y., Lu, X., and Zhang, L., 2024, Occlusion-aware road extraction network for high- resolution remote sensing imagery, IEEE Transactions on Geoscience and Remote Sensing. 62, 5619316.

10.1109/TGRS.2024.3387945
29

Ye, C. S., 2017, Shadow removal from high resolution satellite imagery using vertical borders of shadow region, Journal of Institute of Control, Robotics and Systems, 23(9), 764-770.

10.5302/J.ICROS.2017.17.0132
30

Zhong, Z., Zheng, L., Kang, G., Li, S., and Yang, Y., 2020, Random erasing data augmentation, In Proceedings of the AAAI Conference on Artificial Intelligence, 34(07), 13001-13008.

10.1609/aaai.v34i07.7000
Information
  • Publisher :The Korean Geographical Society
  • Publisher(Ko) :대한지리학회
  • Journal Title :Journal of the Korean Geographical Society
  • Journal Title(Ko) :대한지리학회지
  • Volume : 60
  • No :2
  • Pages :187-196
  • Received Date : 2025-04-04
  • Revised Date : 2025-04-16
  • Accepted Date : 2025-04-17
  • DOI :https://doi.org/10.22776/kgs.2025.60.2.187
Journal Informaiton Journal of the Korean Geographical Society Journal of the Korean Geographical Society
online submission
  • NRF
  • KISTI Current Status
  • KISTI Cited-by
  • KCI 문헌 유사도 검사
  • orcid
  • open access
  • ccl
Journal Informaiton Journal Informaiton - close