基于深度森林的遥感图像变化检测模型

doi:10.6046/zrzyyg.2024327

摘要
图/表
参考文献
相关文章
Metrics

全文: PDF(3458 KB)

HTML
输出: BibTeX | EndNote (RIS)

摘要

针对目前基于深度学习的遥感图像变化检测方法中网络的多粒度性和分类性不高,且对参数敏感需要进行大量调参的问题,该文提出一种基于深度森林的遥感图像变化检测模型。首先采用基础变化检测方法获得初步结果,然后利用深度森林子网络多粒度扫描的特点和强大的数据分类能力对初步结果进行优化获得最终的变化检测结果。选取了多种常见的变化检测模型分别在LEVIR-CD数据集和SYSU-CD数据集上进行验证,并进行了损失函数对比、小样本实验和消融实验,结果表明所提方法在精度、F1得分和召回率等指标上相较于现有经典模型均有显著提高; 所提方法在小数据集上表现出良好的适应性,一定程度上缓解参数调优复杂度和其他深度学习子网络不适用于中小数据集的问题。

	服务

	把本文推荐给朋友
	加入引用管理器
	E-mail Alert
	RSS
	作者相关文章
	葛利华
	王鹏
	张燕琴
	赵双林

关键词 ：遥感, 变化检测, 深度森林, 多粒度扫描, 级联森林

Abstract：

The deep learning-based models currently available for detecting changes in remote sensing images face several challenges, including limited multi-granularity, poor classification performance of networks, high sensitivity to parameters, and great efforts in parameter adjustment. To address these challenges, this study proposed a deep forest-based model for detecting changes in remote sensing images. Initially, preliminary results were determined using a basic change detection method. Then, the results were optimized using the multi-granularity scanning characteristics and strong data classification of deep forest sub-networks. In this manner, the final change detection results were obtained. Verification experiments conducted on the LEVIR-CD and SYSU-CD datasets using various common change detection models indicated that the proposed deep forest-based model significantly outperformed other models in terms of precision, F1 score, and recall. Additionally, the proposed model exhibited strong adaptability on small datasets, as verified by loss function comparison, small-sample experiments, and ablation studies. This adaptability can reduce the complexity of parameter adjustment and address the issues that other deep learning sub-networks fail to be applicable to medium and small datasets.

Key words： remote sensing change detection deep forest multi-granularity scanning cascade forest

收稿日期: 2024-10-07 出版日期: 2025-12-31

ZTFLH:	TP79
	TN912.3

基金资助:国家自然科学基金项目“并行路径支持下的遥感图像超分辨率制图研究”(61801211);河北省资源环境灾变机理及风险监控重点实验室项目“基于多源遥感信息融合的资源环境灾变风险评价与管理”(FZ248201);自然资源部航空地球物理与遥感地质重点实验室项目“超大城市自然资源时空大数据多源融合及其亚像元级制图研究”(2023YFL35);广东省基础与应用基础研究基金项目“面向具有较大变化差异先验图像的时空超分辨率制图研究”(2025A1515010258);深圳市科技计划项目“基于深度学习的机载SAR可视化和文本显示研究”(JCYJ20240813180005007);江苏省自然科学基金项目“高光谱影像智能识别网络特征提取可解释性研究”(BK20221478);湖南省地质灾害监测预警与应急救援工程技术研究中心开放基金项目“基于多源遥感数据的地质灾害识别与监测研究”(hndzgczx202407)

通讯作者: 王鹏(1989-),男,博士,副教授,主要从事遥感信息处理方面的研究。Email: Pengwang_B614080003@nuaa.edu.cn。

作者简介: 葛利华(2001-),男,硕士研究生,主要从事信息获取与处理方面的研究。Email: 042000422@nuaa.edu.cn。

引用本文:

葛利华, 王鹏, 张燕琴, 赵双林. 基于深度森林的遥感图像变化检测模型[J]. 自然资源遥感, 2025, 37(6): 118-127.
GE Lihua, WANG Peng, ZHANG Yanqin, ZHAO Shuanglin. Deep forest-based model for detecting changes in remote sensing images. Remote Sensing for Natural Resources, 2025, 37(6): 118-127.

链接本文:

https://www.gtzyyg.com/CN/10.6046/zrzyyg.2024327 或 https://www.gtzyyg.com/CN/Y2025/V37/I6/118

Fig.1 CDMDF的总体流程图

Fig.2 多粒度扫描示意图

Fig.3 级联森林示意图

Fig.4 LEVIR-CD数据集示例

Fig.5 SYSU-CD数据集示例

Fig.6 LEVIR-CD测试集上不同方法的结果

Fig.7 SYSU-CD测试集上不同方法的结果

Tab.1 2个数据集上各方法的对比

Tab.2 不同损失函数的影响

Fig.8 小样本数据集的结果

Tab.3 数据集的大小对实验结果的影响

Fig.9 2种后处理方法的比较

Tab.4 深度森林对运行时间的影响

[1]	Singh A. Review article digital change detection techniques using remotely-sensed data[J]. International Journal of Remote Sensing, 1989, 10(6):989-1003. doi: 10.1080/01431168908903939
[2]	Huang B, Song H, Cui H, et al. Spatial and spectral image fusion using sparse matrix factorization[J]. IEEE Transactions on Geoscience and Remote Sensing, 2014, 52(3):1693-1704. doi: 10.1109/TGRS.2013.2253612
[3]	Wang P, Wang L, Leung H, et al. Super-resolution mapping based on spatial-spectral correlation for spectral imagery[J]. IEEE Transactions on Geoscience and Remote Sensing, 2021, 59(3):2256-2268. doi: 10.1109/TGRS.36
[4]	Wang Q, Ding X, Tong X, et al. Spatio-temporal spectral unmixing of time-series images[J]. Remote Sensing of Environment, 2021,259:112407.
[5]	Wang P, Yao H, Li C, et al. Multiresolution analysis based on dual-scale regression for pansharpening[J]. IEEE Transactions on Geoscience and Remote Sensing, 2021,60:5406319.
[6]	Marin C, Bovolo F, Bruzzone L. Building change detection in multitemporal very high resolution SAR images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2015, 53(5):2664-2682. doi: 10.1109/TGRS.2014.2363548
[7]	Huang X, Zhang L, Zhu T. Building change detection from multitemporal high-resolution remotely sensed images based on a morphological building index[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2014, 7(1):105-115. doi: 10.1109/JSTARS.4609443
[8]	Mahdavi S, Salehi B, Huang W, et al. A PolSAR change detection index based on neighborhood information for flood mapping[J]. Remote Sensing, 2019, 11(16):1854. doi: 10.3390/rs11161854
[9]	Huang B, Song H. Spatiotemporal reflectance fusion via sparse representation[J]. IEEE Transactions on Geoscience and Remote Sensing, 2012, 50(10):3707-3716. doi: 10.1109/TGRS.2012.2186638
[10]	He D, Zhong Y, Wang X, et al. Deep convolutional neural network framework for subpixel mapping[J]. IEEE Transactions on Geoscience and Remote Sensing, 2021, 59(11):9518-9539. doi: 10.1109/TGRS.2020.3032475
[11]	Gong M, Zhao J, Liu J, et al. Change detection in synthetic aperture Radar images based on deep neural networks[J]. IEEE Transactions on Neural Networks and Learning Systems, 2016, 27(1):125-138. doi: 10.1109/TNNLS.2015.2435783 pmid: 26068879
[12]	史文中, 张鹏林, 吕志勇, 等. 地理国情综合指数及其计算模型研究[J]. 测绘地理信息, 2016, 41(1):1-6.
	Shi W Z, Zhang P L, Lyu Z Y, et al. Composite index and calculation model for national geographical state[J]. Journal of Geoma-tics, 2016, 41(1):1-6.
[13]	李德仁, 眭海刚, 单杰. 论地理国情监测的技术支撑[J]. 武汉大学学报(信息科学版), 2012, 37(5):505-512,502.
	Li D R, Sui H G, Shan J. Discussion on key technologies of geographic national conditions monitoring[J]. Geomatics and Information Science of Wuhan University, 2012, 37(5):505-512,502.
[14]	Chen C F, Son N T, Chang N B, et al. Multi-decadal mangrove forest change detection and prediction in Honduras,Central America,with Landsat imagery and a Markov chain model[J]. Remote Sensing, 2013, 5(12):6408-6426. doi: 10.3390/rs5126408
[15]	Mei S, Li X, Liu X, et al. Hyperspectral image classification using attention-based bidirectional long short-term memory network[J]. IEEE Transactions on Geoscience and Remote Sensing, 2021,60:5509612.
[16]	Adão T, Hruška J, Pádua L, et al. Hyperspectral imaging:A review on UAV-based sensors,data processing and applications for agriculture and forestry[J]. Remote Sensing, 2017, 9(11):1110. doi: 10.3390/rs9111110
[17]	李淑坤, 李培军, 程涛. 加入多时相纹理的遥感变化检测[J]. 国土资源遥感, 2009, 21(3):35-40.doi:10.6046/gtzyyg.2009.03.07.
	Li S K, Li P J, Cheng T. Remote sensing change detection by inclusion of multitemporal texture[J]. Remote Sensing for Land and Resources, 2009, 21(3):35-40.doi:10.6046/gtzyyg.2009.03.07.
[18]	Sivic, Zisserman. Video Google:A text retrieval approach to object matching in videos[C]// Proceedings Ninth IEEE International Conference on Computer Vision. October 13-16,2003, Nice,France.IEEE, 2003:1470-1477.
[19]	Malila W. Change vector analysis:An approach for detecting forest changes with Landsat[C]// LARS symposia.1980:385.
[20]	邸凤萍, 朱重光, 丁玲. 方向矢量法在城市土地利用变化检测中的应用[J]. 计算机工程, 2008, 34(2):253-254. doi: 10.3969/j.issn.1000-3428.2008.02.085
	Di F P, Zhu C G, Ding L. Application of direction-vector analysis in change detection of urban land-use[J]. Computer Engineering, 2008, 34(2):253-254. doi: 10.1007/s00366-017-0537-7
[21]	Fang S, Li K, Shao J, et al. SNUNet-CD:A densely connected Siamese network for change detection of VHR images[J]. IEEE Geoscience and Remote Sensing Letters, 2021,19:8007805.
[22]	Li K, Li Z, Fang S. Siamese NestedUNet networks for change detection of high resolution satellite image[C]// 2020 International Conference on Control,Robotics and Intelligent System.Xiamen China. ACM, 2020:42-48.
[23]	Daudt R C, Le Saux B, Boulch A. Fully convolutional Siamese networks for change detection[C]// 2018 25th IEEE International Conference on Image Processing (ICIP).October 7-10,2018,Athens,Greece.IEEE, 2018:4063-4067.
[24]	Daudt R C. Convolutional neural networks for change analysis in earth observation images with noisy labels and domain shifts[D]. Paris: Institut Polytechnique de Paris, 2020.
[25]	Liu Y, Pang C, Zhan Z, et al. Building change detection for remote sensing images using a dual-task constrained deep Siamese convolutional network model[J]. IEEE Geoscience and Remote Sensing Letters, 2021, 18(5):811-815. doi: 10.1109/LGRS.2020.2988032
[26]	Chen H, Shi Z. A spatial-temporal attention-based method and a new dataset for remote sensing image change detection[J]. Remote Sensing, 2020, 12(10):1662. doi: 10.3390/rs12101662
[27]	Fang H, Du P, Wang X. A novel unsupervised binary change detection method for VHR optical remote sensing imagery over urban areas[J]. International Journal of Applied Earth Observation and Geoinformation, 2022,108:102749.
[28]	杜俊翰, 赖健, 王雪, 等. 基于多尺度注意力特征与孪生判别的遥感影像变化检测及其抗噪性研究[J]. 数据采集与处理, 2022, 37(1):35-48.
	Du J H, Lai J, Wang X, et al. Change detection of remote sensing image based on Siamese multi-scale attention network and its anti-noise ability research[J]. Journal of Data Acquisition and Processing, 2022, 37(1):35-48.
[29]	Wang D, Chen X, Guo N, et al. STCD:Efficient Siamese transformers-based change detection method for remote sensing images[J]. Geo-Spatial Information Science, 2024, 27(4):1192-1211. doi: 10.1080/10095020.2022.2157762
[30]	Zhou Z H, Feng J. Deep forest[J]. National science review, 2019, 6(1):74-86. doi: 10.1093/nsr/nwy108
[31]	杨惠. 基于深度森林的SAR图像变化检测技术研究[D]. 西安: 西安电子科技大学, 2019.
	Yang H. Research on SAR image change detection technology based on deep forest[D]. Xi’an: Xidian University, 2019.
[32]	李恒. 基于遥感影像的森林变化检测方法研究[D]. 长沙: 中南林业科技大学, 2022.
	Li H. Research on forest change detection method based on remote sensing image[D]. Changsha: Central South University of Forestry and Technology, 2022.
[33]	Breiman L. Random forests[J]. Machine Learning, 2001,45:5-32.

[1]	李凯旋, 刘俊伟, 王智博, 蒋文泷, 蔡汉林, 雷少刚, 杨永均. 基于Sentinel-2影像和BenchSegNet模型的露天煤矿台阶提取[J]. 自然资源遥感, 2025, 37(6): 148-155.
[2]	刘晋宇, 胡晋山, 康建荣, 朱益虎, 王胜利. 基于PSR和时序预测模型的露天煤矿区生态环境质量遥感评价[J]. 自然资源遥感, 2025, 37(6): 182-190.
[3]	张莹, 陈运春, 郭晓飞, 吴晓聪, 陈凤林, 曾维军. 基于改进DeepLabV3plus架构的洱海流域水体精细提取[J]. 自然资源遥感, 2025, 37(6): 201-210.
[4]	史晓辰, 罗晨颖, 张超, 王伟, 陈畅, 白雪川, 李少帅. 基于遥感数据的黑土区耕地生态防护评价[J]. 自然资源遥感, 2025, 37(6): 219-227.
[5]	王皓琛, 贺鹏, 陈虹, 童立强, 郭兆成, 涂杰楠, 王根厚. 基于多源遥感的1995—2023年间色林错水域变化特征及淹没趋势分析[J]. 自然资源遥感, 2025, 37(6): 251-262.
[6]	廖远鸿, 白玉琪. 中国区域的国际重要湿地制图产品间一致性分析[J]. 自然资源遥感, 2025, 37(6): 41-48.
[7]	张杰, 王恒友, 霍连志. 基于特征增强和三流Transformer的高光谱遥感图像全色锐化[J]. 自然资源遥感, 2025, 37(6): 97-106.
[8]	何海洋, 李士杰, 秦昊洋, 刘小玉, 王思琪, 孙旭. 基于资源一号02D高光谱数据的甘肃北山前红泉地区蚀变矿物填图及特征分析[J]. 自然资源遥感, 2025, 37(5): 195-205.
[9]	郎云帆, 李益敏, 李媛婷, 刘淼, 柏克冰. 基于夜光遥感的滇缅地区经济发展时空特征研究[J]. 自然资源遥感, 2025, 37(5): 233-242.
[10]	邹娟, 余姝辰, 贺秋华, 徐质彬, 尹向红, 邹聪. 1933—2024年洞庭湖注滋口三角洲时空演变特征[J]. 自然资源遥感, 2025, 37(5): 24-31.
[11]	张萍, 庞咏, 陈庆松, 杨坤, 邹祖建, 侯云花, 王彩琼, 冯思齐. 基于RSEI模型的滇西北高山峡谷区生态环境质量动态监测及驱动因子分析[J]. 自然资源遥感, 2025, 37(5): 243-253.
[12]	杨赈, 杨明龙, 李国柱, 夏永华, 俞婷, 严正飞, 李万涛. 修正水分胁迫的NPP反演结果与典型高原盆地土壤水分关系探究[J]. 自然资源遥感, 2025, 37(5): 267-277.
[13]	何威荣, 霍润斌, 李东辉, 张峰山, 王新峰. 基于GIS的安徽省自然湖泊动态变化及驱动力分析[J]. 自然资源遥感, 2025, 37(5): 32-43.
[14]	龚绍军, 陈超, 范竞. 一种考虑地物含水量和长时序遥感影像的海岸线精准定位方法[J]. 自然资源遥感, 2025, 37(5): 53-61.
[15]	吴志军, 丛铭, 许妙忠, 韩玲, 崔建军, 赵超英, 席江波, 杨成生, 丁明涛, 任超锋, 顾俊凯, 彭晓东, 陶翊婷. 基于视觉双驱动认知的高分辨率遥感影像自学习分割方法[J]. 自然资源遥感, 2025, 37(5): 73-90.

Viewed

Full text

Abstract

Cited

Shared

Discussed