Landslide Remote Sensing Image Recognition Based on EfficientNet: Taking Bijie City, Guizhou Province as an Example
-
摘要: 近年来,随着工程建设的快速发展,工程活动改变了边坡原始地质条件,导致滑坡灾害频繁发生,严重威胁人民的生命财产安全。因此,深入研究滑坡的快速、精确识别方法对于防灾减灾具有重要意义。本文提出一种基于EfficientNet 高效网络提取滑坡深度特征的潜在滑坡识别方法,该方法通过寻找一组最优的复合系数从深度、宽度、分辨率三个维度对神经网络进行扩展,自适应地优化网络结构,并引入带动量的梯度下降算法(Stochastic Gradient Descent Momentum,SGDM)作为网络学习的优化器,充分考虑历史梯度的影响,在参数更新过程中不断调整当前梯度值,从而相应地调整参数的更新幅度,改善神经网络的学习效果,提取滑坡体的深层次特征。实验结果表明,EfficientNet 模型在测试集上的平均准确度达到92.78%,可以高效准确地实时提取滑坡信息,对灾后的快速反应有指导意义。
-
关键词:
- 滑坡识别 /
- 深度特征 /
- EfficientNet /
- 带动量的梯度下降算法(SGDM)
Abstract: In recent years, with the rapid development of engineering construction, engineering activities have changed the original geological conditions of slopes, resulting in frequent landslide disasters, which seriously threaten people's life and property safety. Therefore, it is of great significance to study the rapid and accurate identification method of landslides for disaster prevention and reduction. In this paper, a potential landslide recognition method is proposed based on EfficientNet to realize the extraction of landslide depth features. The method extends the neural network from three dimensions of depth, width, and resolution by searching for a set of optimal composite coefficients, and adaptively optimizes the network structure. The Stochastic Gradient Descent Momentum (SGDM) is introduced as the optimizer of network learning, which fully considers the influence of historical gradient. And the current gradient value is constantly adjusted during the param eter updating process, so as to adjust the parameter updating amplitude accordingly, improve the learning effect of neural networks and extract the deep features of the slope. The experimental results show that the average accuracy of the EfficientNet model on the test set reaches 92.78%, which can efficiently and accurately extract landslide information in real-time and provides guiding information for the rapid response after the disaster. -
[1] 陈聪,候磊,李乐乐,杨鑫涛.2021.基于GRU改进RNN神经网络的飞机燃油流量预测[J].科学技术与工程,21(27):11663-11673.
[2] 陈冠宇,李向,王岭玲.2016a.基于大数据的遥感图像植被识别方法[J].地质科技情报,35(3):204-209.
[3] 陈冠宇,安凯,李向.2016b.基于卷积神经网络的不良地质体识别与分类[J].地质科技情报,35(1):205-211.
[4] 陈慧娟,邹浩,訚遥,王超.2023.持续强降雨影响下黄梅县袁山村三组滑坡破坏特征与成因分析[J].华南地质,39(3):482-491.
[5] 姜英, 王延江, 林青, 刘伟锋.2017. 基于卷积神经网络与Bayesian 决策的图像识别与分类记忆建模[J].中国科学:技术科学,47(9):977-984.
[6] 康牧,凌凤彩.2013.一种基于新插值方法的图像旋转算法[J].计算机科学,40(5):303-306.
[7] 刘沛,况顺达,姚智.2004.RS、GIS 在贵州省环境地质综合调查中的应用——以崩塌、滑坡、泥石流调查为例[J].贵州地质,21(3):161-164+160.
[8] 李剑萍.2004.3S技术在灾害监测预测中的应用及展望[J].灾害学,19(S1):85-89.
[9] 卢薇艳,罗鹏,龚淑云.2019.基于遥感技术的海岸线提取及应用研究综述[J].华南地质与矿产,35(3):393-397.
[10] 李长冬,龙晶晶,姜茜慧,付智勇.2020.水库滑坡成因机制研究进展与展望[J].地质科技通报,39(1):67-77.
[11] 李长冬,谭钦文.2022.动水驱动型滑坡物理启滑能够预测吗?[J].地球科学,47(10):3908-3910.
[12] 李长冬,孟杰,项林语,黄徳崴,崔宇寒.2022.白鹤滩库首区砂岩结构多尺度演变机制研究[J/OL].地球科学:1-10.
[13] 李益敏,袁静,蒋德明,王东驰,刘心知.2021.基于GIS 的高山峡谷区滑坡灾害危险性评价——以泸水市为例[J].水土保持研究,28(3):355-363.
[14] 马玉林,吴华,张文海.2022.遥感技术在滑坡识别中的应用探讨[J].科技创新与应用,12(3):142-145.
[15] 邱芹军,吴亮,马凯,谢忠,陶留锋.2023.面向灾害应急响应的地质灾害链知识图谱构建方法[J].地球科学,48(5):1875-1891.
[16] 唐辉明,李长冬,龚文平,邹宗兴,张永权,张抒,张俊荣.2022.滑坡演化的基本属性与研究途径[J].地球科学,47(12):4596-4608.
[17] 唐辉明.2022.重大滑坡预测预报研究进展与展望[J].地质科技通报,41(6):1-13.
[18] 徐宏根,李春来,杨淼,董小环.2019.光谱保真归一化卷积高光谱超分辨率重建方法[J]. 华南地质与矿产,35(4):504-513.
[19] 杨春雨,张鑫.2022.煤矿机器人环境感知与路径规划关键技术[J].煤炭学报,47(7):2844-2872.
[20] 殷宗敏,赵宝强,叶润青.2022.时序InSAR技术在三峡库首区潜在滑坡识别中的应用研究[J]. 华南地质,38(2):273-280.
[21] 周乐群,杨岚.2000.基于“3S”技术的国土资源与生态环境动态监测[J].华南地质与矿产,16(4):40-46.
[22] 周前祥,敬忠良,姜世忠.2004.遥感影像双线性插值小波融合方法[J].上海交通大学学报,38(4):547-550.
[23] 张勤,黄观文,杨成生.2017.地质灾害监测预警中的精密空间对地观测技术[J].测绘学报,46(10):1300-1307.
[24] 张勤,赵超英,陈雪蓉.2022.多源遥感地质灾害早期识别技术进展与发展趋势[J].测绘学报,51(6):885-896.
[25] 张芳,赵东旭,肖志涛,耿磊,吴骏,刘彦北.2022.单幅图像超分辨率重建技术研究进展[J]. 自动化学报,48(11):2634-2654.
[26] 张占忠,陈富强,刘亚林,李丹,杜世回,张晓宇.2023.基于遥感技术的嘉黎江巴变形体稳定性评价与分析[J].地质科技通报,42(3):28-37.
[27] Baatz M. 1999. Object-oriented and multi-scale image analysis in semantic networks[C].Proc the International Symposium on Operationali-zation of Remote Sensing.
[28] Behling R, Roessner S, Kaufmann H, Kleinschmit B. 2014. Automated Spatiotemporal Landslide Mapping over Large Areas Using RapidEye Time Series Data[J]. Remote Sensing, 6(9): 8026-8055.
[29] Blesius L J. 2002. A satellite based methodology for landslide susceptibility map-ping incorporating geotechnical slope stability parameters. [D]. The University of Iowa.
[30] Chen L, Lin W B, Chen P, Jiang S, Liu L, Hu H Y. 2021. Porosity Prediction from Well Logs Using Back Propagation Neural Network Optimized by Genetic Algorithm in One Heterogeneous Oil Reservoirs of Ordos Basin[J]. Journal of Earth Science, 32(4): 828-838.
[31] Criss R E, Yao W M, Li C D, Tang H M. 2020. A Predictive, Two-Parameter Model for the Movement of Reservoir Landslides[J]. Journal of Earth Science, 31(6): 1051-1057.
[32] Ji S P, Yu D W, Shen C Y, Li W L, Xu Q. 2020. Landslide detection from an open satellite imagery and digital elevation model dataset using attention boosted convolutional neural networks[J]. Landslides, 17: 1337-1352.
[33] Krizhevsky A, Sutskever I, Hinton G E. 2017. ImageNet classification with deep convolutional neural networks [J]. Communications of the ACM, 60(6), 84-90.
[34] Li S, Chen J P, Liu C, Wang Y. 2021. Mineral Prospectivity Prediction via Convolutional Neural Networks Based on Geological Big Data[J]. Journal of Earth Science, 32(2): 327-347.
[35] Long Y J, Li W L, Huang R Q, Xu Q, Yu B, Liu G. 2023. A Comparative Study of Supervised Classification Methods for Investigating Landslide Evolution in the Mianyuan River Basin, China[J]. Journal of Earth Science, 34(2): 316-329.
[36] Mantovani F, Soeters R, Westen C. 1996. Remote sensing techniques for landslide studies and hazard zonation in Europe[J]. Geomorphology, 15(3): 213-225.
[37] Mezaal M R, Pradhan B. 2018. An improved algorithm for identifying shallow and deep-seated landslides in dense tropical forest from airborne laser scanning data[J]. Catena, 167: 147-159.
[38] Noferini L, Pieraccini M, Mecatti D, Macaluso G, Atzeni C, Mantovani M, Marcato G, Pasuto A, Silvano S, Tagliavini F. 2007. Using GB-SAR technique to monitor slow moving landslide[J]. Engineering Geology, 95(3): 88-98.
[39] Roessner S, Wetzel H U, Kaufmann H, Sarnagoev A. 2005. Potential of Satellite Remote Sensing and GIS for Landslide Hazard Assessment in Southern Kyrgyzstan (Central Asia) [J]. Natural Hazards, 35(3): 395-416.
[40] Schulz W H.2007.Landslide susceptibility revealed by LIDAR imagery and historical records, Seattle, Washington[J]. Engineering Geology, 89(1-2): 67-87.
[41] Schuster R L, Krizek R. 1979. Landslides: Analysis and control[J]. Transportation Research Board Special Report.
[42] Stephens P R, Trotter C M, De Rose R C, Newsome P F, Carr K S. 1988. Use of satellite data to map landslides[C]. 9th Asian Conference on Remote Sensing, Bangkok, Thailand, J11.1-J11.7.
[43] Su A J, Feng M Q, Dong S, Zou Z X, Wang J G. 2022. Improved Statically Solvable Slice Method for Slope Stability Analysis[J]. Journal of Earth Science, 33(5): 1190-1203.
[44] Tan M X, Le Q V. 2019. EfficientNet: Re-thinking Model Scaling for Convolutional Neural Networks[C]. 36th International Conference on Machine Learning (ICML).
[45] Zhong C, Li H, Xiang W, Su A J, Huang X F. 2012. Comprehensive Study of Landslides through the Integration of Multi Remote Sensing Techniques: Framework and Latest Advances[J]. Journal of Earth Science, 23(2): 243-252.
计量
- 文章访问数: 408
- PDF下载数: 17
- 施引文献: 0