级联多注意力的遥感影像耕地变化检测方法

Cascaded Multi-attention Method for Detecting Cropland Changes in Remote Sensing Images

摘要

针对当前深度卷积神经网络在处理耕地精细变化识别时,因忽略了特征通道间的相关性以及特征重要性差异导致识别困难的问题,提出了一种新的多注意力遥感影像耕地变化检测网络(Multi-Attention Cropland Change Detection Network,MACCDNet).MACCDNet首先提取双时相高空间分辨率遥感影像中的多尺度耕地特征;然后级联门控通道转换(Gated Channel Transformation,GCT)和挤压-激励(Squeeze-and-Excitation,SE)模块的解码器逐层处理多尺度双时相耕地特征,GCT模块全面建模上下文信息,SE模块对重要性不同的特征赋予相应的关注度,有效提升MACCDNet对耕地变化的识别能力;最后通过融合多尺度特征和图像差异特征得到耕地变化结果.MACCDNet在CLCD和JL-1耕地变化监测数据集上与6种先进方法相比,综合性能最优.结果表明,MACCDNet是一种精确、高效和高鲁棒性的耕地变化检测方法.

Abstract

Aiming at the problem that the existing deep convolutional neural network ignores the channel correlation and importance difference between different features,which leads to the difficulty of identifying fine cropland changes.This paper proposes a new multi-attention crop change detection network (MACCDNet) for remote sensing images.Firstly,MACCDNet extracts multi-scale cropland features from bi-temporal high spatial resolution remote sensing images.Then,the decoder of gated channel transformation (GCT) and squeee-and-excitation (SE) module are cascaded to process multi-scale and bi-temporal cropland features layer by layer.The GCT module comprehensively model’s context information.The SE module gives corresponding attention to features with different importance,which effectively improves the recognition ability of MACCDNet for cropland change.Finally,the cropland change results are obtained by fusing multi-scale features and image difference features.MACCDNet has the best comprehensive performance compared with six advanced methods on the CLCD and JL-1 cropland change detection datasets.The results show that MACCDNet is an accurate,efficient,and robust method for cultivated land change detection.

关键词

耕地变化监测 / 高分辨率遥感影像 / 耕地保护 / 卷积神经网络 / 级联多注意力

Key words

cropland change detection / high resolution remote sensing image / cropland protection / convolutional neural network / cascade multi-attention

参考文献

[1] 眭海刚, 王建勋, 华丽, 等. 遥感耕地监测现状与方法综述[J]. 广西科学, 2022, 29(1):1-12.

[2] SUI H G, WANG J X, HUA L, et al. Review on the status and methods of remote sensing farmland monitoring[J]. Guangxi Sciences, 2022, 29(1):1-12.

[3] 周江明. 中国耕地重金属污染现状及其人为污染源浅析[J]. 中国土壤与肥料, 2020(2):83-92.

[4] ZHOU J M. The present status of heavy metal(loid)s pollution in farmland soils and analysis of polluting sources in China[J]. Soils and Fertilizers Sciences in China, 2020(2):83-92.

[5] 张祖勋, 姜慧伟, 庞世燕, 等. 多时相遥感影像的变化检测研究现状与展望[J]. 测绘学报, 2022, 51(7):1091-1107.

[6] ZHANG Z X, JIANG H W, PANG S Y, et al. Review and prospect in change detection of multi-temporal remote sensing images[J]. Acta Geodaetica et Cartographica Sinica, 2022, 51(7):1091-1107.

[7] XU Y D, YU L, ZHAO F R, et al. Tracking annual cropland changes from 1984 to 2016 using time-series Landsat images with a change-detection and post-classification approach:Experiments from three sites in Africa[J]. Remote Sensing of Environment, 2018, 218:13-31.

[8] MARDIAN J, BERG A, DANESHFAR B. Evaluating the temporal accuracy of grassland to cropland change detection using multitemporal image analysis[J]. Remote Sensing of Environment, 2021, 255:112292.

[9] WANG Z H, YAO W Y, TANG Q H, et al. Continuous change detection of forest/grassland and cropland in the Loess Plateau of China using all available landsat data[J]. Remote Sensing, 2018, 10(11):1775.

[10] JI C Y. Delineating agricultural field boundaries from TM imagery using dyadic wavelet transforms[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 1996, 51(6):268-283.

[11] WU Q, HUANG L, TANG B H, et al. CroplandCDNet:Cropland change detection network for multitemporal remote sensing images based on multilayer feature transmission fusion of an adaptive receptive field[J]. Remote Sensing, 2024, 16(6):1061.

[12] XU C, YE Z Y, MEI L Y, et al. Cross-attention guided group aggregation network for cropland change detection[J]. IEEE Sensors Journal, 2023, 23(12):13680-13691.

[13] LIU M X, CHAI Z Q, DENG H J, et al. A CNN-transformer network with multiscale context aggregation for fine-grained cropland change detection[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2022, 15:4297-4306.

[14] WU Z H, CHEN Y, MENG X L, et al. SwinUCDNet:A UNet-like network with union attention for cropland change detection of aerial images[C]// 2023 30th International Conference on Geoinformatics,19-21 July 2023, London,United Kingdom.IEEE,2023:1-7.

[15] YANG Z X, ZHU L C, WU Y, et al. Gated channel transformation for visual recognition[C]// 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR),13-19 June 2020, Seattle,WA,USA.IEEE,2020:11791-11800.

[16] HU J, SHEN L, SUN G. Squeeze-and-excitation networks[C]// 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition,18-23 June 2018, Salt Lake City,UT,USA.IEEE,2018:7132-7141.

[17] SIMONYAN K, ZISSERMAN A. Very deep convolutional networks for large-scale image recognition[EB/OL]. 2014:arXiv: 1409.1556.https://arxiv.org/abs/1409.1556.

[18] ZHANG J D, SHAO Z F, DING Q, et al. AERNet:An attention-guided edge refinement network and a dataset for remote sensing building change detection[J]. IEEE Transactions on Geoscience and Remote Sensing, 2023, 61:5617116.

[19] DE BEM P P, DE CARVALHO O A Jr, DE CARVALHO O L F, et al. Performance analysis of deep convolutional autoencoders with different patch sizes for change detection from burnt areas[J]. Remote Sensing, 2020, 12(16):2576.

[20] PAPADOMANOLAKI M, VAKALOPOULOU M, KARANTZALOS K. A deep multitask learning framework coupling semantic segmentation and fully convolutional LSTM networks for urban change detection[J]. IEEE Transactions on Geoscience and Remote Sensing, 2021, 59(9):7651-7668.

[21] CHEN H, QI Z P, SHI Z W. Remote sensing image change detection with transformers[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60:5607514.

[22] ALCANTARILLA P F, STENT S, ROS G, et al. Street-view change detection with deconvolutional networks[J]. Autonomous Robots, 2018, 42(7):1301-1322.

[23] LIN M H, YANG G Y, ZHANG H Y. Transition is a process:Pair-to-video change detection networks for very high resolution remote sensing images[J]. IEEE Transactions on Image Processing, 2023, 32:57-71.

[24] FANG S, LI K Y, SHAO J Y, et al. SNUNet-CD:A densely connected Siamese network for change detection of VHR images[J]. IEEE Geoscience and Remote Sensing Letters, 2022, 19:8007805.

[25] KINGMA D P, BA J. Adam: A method for stochastic optimization[EB/OL]. 2014:arXiv: 1412.6980.https://arxiv.org/abs/1412.6980.

[26] 张良培, 武辰. 多时相遥感影像变化检测的现状与展望[J]. 测绘学报, 2017, 46(10):1447-1459.

[27] ZHANG L P, WU C. Advance and future development of change detection for multi-temporal remote sensing imagery[J]. Acta Geodaetica et Cartographica Sinica, 2017, 46(10):1447-1459.

基金资助

湖南省自然科学基金项目(2024JJ8317)

国家自然科学基金项目(42361054)

湖南省地质院科技计划项目(HNGSTP202409)

云南省“兴滇英才支持计划”项目(KKRD202221036)