计算机科学

计算机科学 ›› 2024, Vol. 51 ›› Issue (11A): 240300095-8.doi: 10.11896/jsjkx.240300095

• 图像处理&多媒体技术 • 上一篇    下一篇

基于多任务学习的复杂城市遥感图像道路提取

王坤阳1, 刘洋1, 业宁1, 张凯2   

  1. 1 南京林业大学信息科学技术学院 南京 210037
    2 东南大学仪器科学与工程学院 南京 210018
  • 出版日期:2024-11-16 发布日期:2024-11-13
  • 通讯作者: 业宁(yening@njfu.edu.cn)
  • 作者简介:(1961198686@qq.com)
  • 基金资助:
    未来网络科研基金项目(FNSRFP-2021-YB-17)

Road Extraction from Complex Urban Remote Sensing Images Based on Multi-task Learning

WANG Kunyang1, LIU Yang1, YE Ning1, ZHANG Kai2   

  1. 1 School of Information Science and Technology,,Nanjing Forestry University,Nanjing 210037,China
    2 School of Instrument Science and Engineering,Southeast University,Nanjing 210018,China
  • Online:2024-11-16 Published:2024-11-13
  • About author:WANG Kunyang,born in 1993,M.S.His main research interests include embedded technology and so on.
    YE Ning,born in 1967,doctor,professor.His main research interests include bioinformatics,data mining and machine learning.
  • Supported by:
    Future Network Research Fund Program (FNSRFP-2021-YB-17).

PDF (PC)

147

摘要： 提出一种新的遥感图像道路提取框架,旨在利用从道路边缘检测中获得的知识来提高道路提取的准确性。研究中引入了一个融合多尺度信息和视觉注意力机制的多尺度视觉注意力模块,并构建了一个级联特征融合模块以集成网络在不同尺度上的预测结果。在此基础上,构建了一个包含编码器和解码器的多尺度视觉注意网络(MSVANet)。同时,提出一个多任务学习框架,该框架结合了MSVANet,并采用粒子群优化算法对多任务学习框架的两个学习率超参数的自动选取进行优化。RNBD数据集的训练和测试结果表明,所提方法在各种分割精度指标和泛化能力方面均优于其他道路提取方法。

关键词: 深度学习, 遥感图像道路提取, 多任务学习, 多尺度视觉注意力网络(MSVANet)

Abstract: In this paper,we propose a new framework for road extraction from remote sensing images that aims to utilize the knowledge gained from road edge detection to improve the accuracy of road extraction.A multi-scale visual attention module that fuses multi-scale information and visual attention mechanisms is introduced in the study,and a cascading feature fusion module is constructed to integrate the network's prediction results at different scales.Based on this,we construct a multiscale visual attention network(MSVANet) containing encoders and decoders.A multi-task learning framework that incorporates the MSVANet is also proposed,and a particle swarm optimization algorithm(PSO) is used to optimize the automatic selection of the two learning rate hyperparameters of the multi-task learning framework.The training and testing results on the RNBD dataset show that the proposed method outperforms other road extraction methods in terms of various segmentation accuracy metrics and generalization ability.

Key words: Deep learning, Remote sensing image road extraction, Multi-task learning, Multi-scale visual attention network

中图分类号: 

  • TP391
[1]FANG H W.Application of Telematics in Intelligent Urban Transportation Networks[J].Electronic Technology,2022,51(2):228-229.
[2]MASSA F,BONAMINI L,SETTIMI A,et al.LiDAR-based GNSS Denied Localization for Autonomous Racing Cars[J].Sensors,2020,20(14):3992.
[3]ZHU Z X.Analysis of urban planning and sustainable development in the system of territorial spatial planning[J].Future City Design and Operation,2024(3):20-22.
[4]XU L Y,MAO K B,GUO Z H,et al.A review on the application of convolutional neural network in semantic segmentation of agricultural remote sensing images[J].Agricultural Outlook,2024,20(2):70-75.
[5]ZHU J Z.Research and application of remote sensing image semantic segmentation technology based on deep learning[J].Value Engineering,2023,42(34):109-111.
[6]WANG Z,XIE C M.Application of drone remote sensing technology in agriculture[J].Agricultural Engineering Technology,2023,43(29):44-45.
[7]LIAN R B,ZHANG Z M,LIAO Y P,et al.Fast road centerline extraction from remote sensing images by combining geodetic distance field and curve smoothing[J].Journal of Surveying and Mapping,2023,52(8):1317-1329.
[8]HUANG W K,TENG F,WANG Z D,et al.A review of image segmentation based on deep learning[J].Computer Science,2024,51(2):107-116.
[9]GUO D Q,FU Y,ZHU Y,et al.Self-attentive multi-scale feature fusion algorithm for semantic segmentation of remote sensing images[J].Journal of Computer Aided Design and Graphics,2023,35(8):1259-1268.
[10]MALAMBO L,POPESCU S C,ROONEY W,et al.A DeepLearning Semantic Segmentation-Based Approach for Field-LevelSorghum Panicle Counting [J].Remote Sensing,2019,11(24).
[11]LI X,JIAO H,WANG Y.Edge detection algorithm of cancer image based on deep learning [J].Bioengineered,2020,11(1):693-707.
[12]HUI H,ZHANG X,LI F,et al.A Partitioning-Stacking Prediction Fusion Network Based on an Improved Attention U-Net for Stroke Lesion Segmentation [J].IEEE Access,2020,PP(99):1-1.
[13]GAO D,CHEN J Y,XIE Y.A-PSPNet:APSPNet image semantic segmentation model incorporating attention mechanism[J].Journal of Chinese Academy of Electronic Science,2020,15(6):518-523.
[14]KEMKER R,SALVAGGIO C,KANAN C.Algorithms for semantic segmentation of multispectral remote sensing imagery using deep learning [J].ISPRS Journal of Photogrammetry and Remote Sensing,2018,145(NOV.):60-77.
[15]LIANG Z Y,WANG X L.Semantic segmentation of multispectral remote sensing images based on band-position adaptive selection[J].Advances in Lasers and Optoelectronics,2023,60(14):157-167.
[16]LU Z,QI L,LI B Y.Rice yield classification by combining hyperspectral features with semantic segmentation[J].Computer and Digital Engineering,2023,51(9):1979-1984+2097.
[17]LIU B,YU X,YU A,et al.Deep Few-Shot Learning for Hyperspectral Image Classification [J].IEEE Transactions on Geoscience & Remote Sensing,2018,57(4):2290-2304.
[18]LAKE B M,ULLMAN T D,TENENBAUM J B,et al.Building Machines That Learn and Think Like People [J].Behavioral and Brain Sciences,2017,40:e253.
[19]HOCHREITER S,YOUNGER A S,CONWELL P R.Learning to learn using gradient descent[C]//Artificial Neural Networks-ICANN 2001:International Conference Vienna,Austria,August 21-25,2001 Proceedings 11.Springer Berlin Heidelberg,2001:87-94.
[20]JÜRGEN SCHMIDHUBER,ZHAO J,WIERING M.ShiftingInductive Bias with Success-Story Algorithm,Adaptive Levin Search,and Incremental Self-Improvement [J].Machine Learning,1997,28(1):105-130.
[21]GRAVES A,WAYNE G,DANIHELKA I.Neural Turing Machines [J].arXiv:1410.5401,2014.
[22]XIE S,TU Z.Holistically-Nested Edge Detection[C]// 2015 IEEE International Conference on Computer Vision(ICCV).IEEE,2016.
[23]DOSOVITSKIY A,BEYER L,KOLESNIKOV A,et al.AnImage is Worth 16x16 Words:Transformers for Image Recognition at Scale[C]// International Conference on Learning Representations.2021.
[24]SHIVERS S,ROBERTS D,MCFADDEN J,et al.Using ImagingSpectrometry to Study Changes in Crop Area in California's Central Valley during Drought [J].Remote Sensing,2018,10(10).
[25]LIU Y,YAO J,LU X,et al.RoadNet:Learning to Comprehensively Analyze Road Networks in Complex Urban Scenes From High-Resolution Remotely Sensed Images [J].IEEE Transactions on Geoscience and Remote Sensing,2019,57(4):2043-2056.
[26]FETTAKA S,THIBAULT J,GUPTA Y.A new algorithmusing front prediction and NSGA-II for solving two and three-objective optimization problems [J].Optimization & Engineering,2015,16(4):713-736.
[27]PASZKE,ADAM,et al.Pytorch:An imperative style,high-performance deep learning library [J].Advances in Neural Information Processing Systems,2019,32.
[28]ZHOU L,ZHANG C,WU M.D-LinkNet:LinkNet with pre-trained encoder and dilated convolution for high resolution satellite imagery road extraction[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops.2018:182-186.
[29]WU M,ZHANG C,LIU J,et al.Towards accurate high resolution satellite image semantic segmentation[J].Ieee Access,2019,7:55609-55619.
[30]YAN H,ZHANG C,YANG J,et al.Did-LinkNet:Polishing D-block with dense connection and iterative fusion for road extraction[C]//2021 IEEE International Geoscience and Remote Sensing Symposium IGARSS.IEEE,2021:2186-2189.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
渝ICP备16013121号-4
地址：重庆市渝北区洪湖西路18号    邮编：401121  
电话：023-63500828（编辑部） 023-67039612（会议/专题） 023-67039623（财务/发票）
E-mail：jsjkx12@163.com
本系统由北京玛格泰克科技发展有限公司设计开发