基于多尺度语义信息增强的农田地块提取网络

doi:10.16180/j.cnki.issn1007-7820.2023.07.010

摘要/Abstract

摘要：

针对因农田类间异质性高、相邻地块之间间隔小而出现的相邻地块黏连、地块提取不完整等问题,文中提出了一种基于多尺度语义信息增强的农田地块提取网络。采用并行的多尺度特征提取模块,通过保持特征图的高分辨率保留高精度的边缘信息,降低由下采样带来的细节损耗,缓解地块之间的黏连。使用基于注意力机制的全局语义信息增强模块,通过获取全局语义信息增强网络的类别判断能力减少地块提取不完整的现象。实验结果表明,在IoU、OA、F1-score评价指标上,文中方法比现有研究中具有代表性的4种算法提高了1%~13%。

关键词: 高分辨率遥感影像, 农田地块, 深度学习, 语义分割, 多尺度特征, 语义信息增强, 地块黏连, 地块不完整

Abstract:

Facing the problems of adhesion of adjacent parcels and incomplete parcels due to the high heterogeneity and the small region among neighbor parcels, a farmland parcel extraction network based on multi-scale semantic information enhancement is proposed in this study. To alleviate the adhesion between parcels, the multi-scale feature extraction module with parallel structure is used, which retained the high-resolution feature maps to maintain high-precision edge information and reduce the loss of spatial location information due to downsampling. Furthermore, to reduce the phenomenon of incomplete parcels, the global semantic information enhancement module based on attentional mechanism is utilized to enhance the classification ability of the network by capture global semantic information instead of local semantic information. According to the experimental results, it is shown that the proposed method is 1%~13% better than the four typical algorithms in existing studies in terms of IoU, OA, and F1-score evaluation indexes.

Key words: high-resolution remote sensing imagery, farmland parcel, deep learning, semantic segmentation, multi-scale features, semantic information enhancement, adhesion of adjacent parcel, incomplete parcel

中图分类号:

TP751

曾薪鑫,张洪艳. 基于多尺度语义信息增强的农田地块提取网络[J]. 电子科技, 2023, 36(7): 70-74.

ZENG Xinxin,ZHANG Hongyan. A Farmland Parcel Extraction Network Based on Multi-Scale Semantic Information Enhancement[J]. Electronic Science and Technology, 2023, 36(7): 70-74.

图/表 5

图1

图2

图3

表1

图4

参考文献 20

[1]	Persello C, Tolpekin V A, Bergado J R, et al. Delineation of agricultural fields in smallholder farms from satellite images using fully convolutional networks and combinatorial grouping[J]. Remote Sensing of Environment, 2019, 23(1):1-18. doi: 10.1016/0034-4257(87)90067-8
[2]	Meier J, Mauser W, Hank T, et al. Assessments on the impact of high-resolution-sensor pixel sizes for common agricultural policy and smart farming services in European regions[J]. Computers and Electronics in Agriculture, 2020, 16(9):1-11. doi: 10.1016/S0168-1699(96)00018-X
[3]	Zhang D, Pan Y, Zhang J, et al. A generalized approach based on convolutional neural networks for large area cropland mapping at very high resolution[J]. Remote Sensing of Environment, 2020, 24(7):1-23. doi: 10.1016/0034-4257(88)90003-X
[4]	Turker M, Kok E H. Field-based sub-boundary extraction fromremote sensing imagery using perceptual grouping[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2013, 79(3):106-21. doi: 10.1016/j.isprsjprs.2013.02.009
[5]	Rahman M S, Di L, Yu Z, et al. Crop field boundary delineation using historical crop rotation pattern[C]. Istanbul: Proceedings of the Eighth International Conference on Agro-Geoinformatics,IEEE, 2019:1-5.
[6]	Xue Y, Zhao J, Zhang M. A watershed-segmentation-based improved algorithm for extracting cultivated land boundaries[J]. Remote Sensing, 2021, 13(5):939-943. doi: 10.3390/rs13050939
[7]	Ghaffarian S, Turker M. An improved cluster-based snake model for automatic agricultural field boundary extraction from high spatial resolution imagery[J]. International Journal of Remote Sensing, 2018, 40(4):1217-1247. doi: 10.1080/01431161.2018.1524178
[8]	García-Pedrero A, Gonzalo-Martín C, Lillo-Saavedra M. A machine learning approach for agricultural parcel delineation through agglomerative segmentation[J]. International Journal of Remote Sensing, 2017, 38(7): 1809-1819. doi: 10.1080/01431161.2016.1278312
[9]	Zhang P, Hu S, Li W, et al. Parcel-level mapping of crops in a smallholder agricultural area: A case of central China using single-temporal VHSR imagery[J]. Computers and Electronics in Agriculture, 2020, 17(5):1-11. doi: 10.1016/S0168-1699(96)01229-X
[10]	Yoon H, Kim S. Detecting abandoned farmland using harmonic analysis and machine learning[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2020, 16(6):201-212.
[11]	闫超, 孙占全, 田恩刚, 等. 基于深度学习的医学图像分割技术研究进展[J]. 电子科技, 2021, 34(2):7-11.
	Yan Chao, Sun Zhanquan, Tian Engang, et al. Research progress of medical image segmentation based on deep learning[J]. Electronic Science and Technology, 2021, 34(2):7-11.
[12]	Du Z, Yang J, Ou C, et al. Smallholder crop area mapped with a semantic segmentation deep learning method[J]. Remote Sensing, 2019, 11(7):1-21. doi: 10.3390/rs11010001
[13]	周楠, 杨鹏, 魏春山, 等. 地块尺度的山区耕地精准提取方法[J]. 农业工程学报, 2021, 37(19):260-266.
	Zhou Nan, Yang Peng, Wei Chunshan, et al. Accurate extraction method for cropland in mountainous areas based on field parcel[J]. Transactions of the Chinese Society of Agricultural Engineering, 2021, 37(19):260-266.
[14]	Xia L, Luo J, Sun Y, et al. Deep extraction of cropland parcels from very high-resolution remotely sensed imagery[C].Hangzhou: Proceedings of the Seventh International Conference on Agro-geoinformatics,IEEE, 2018:1-5.
[15]	Sun K, Xiao B, Liu D, et al. Deep high-resolution representation learning for human pose estimation[C]. Long Beach: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019:5693-5703.
[16]	Wang X, Girshick R, Gupta A, et al. Non-local neural networks[C].Salt Lake City: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018:7794-7803.
[17]	Zhu Q, Liao C, Hu H, et al. MAP-Net: Multiple attending path neural network for building footprint extraction from remote sensed imagery[J]. IEEE Transactions on Geoscience and Remote Sensing, 2020, 59(7):6169-6181. doi: 10.1109/TGRS.2020.3026051
[18]	Shelhamer E, Long J, Darrell T. Fully convolutional networks for semantic segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 39(4):640-651. doi: 10.1109/TPAMI.2016.2572683
[19]	Ronneberger O, Fischer P, Brox T. U-net: Convolutional networks for biomedical image segmentation[C]. Freiburg: International Conference on Medical Image Computing and Computer-assisted Intervention,Springer Cham, 2015:234-241.
[20]	Chen L C, Zhu Y, Papandreou G, et al. Encoder-decoder with atrous separable convolution for semantic image segmentation[C].Munich: Proceedings of the European Conference on Computer Vision, 2018:801-818.

算法	IoU	OA	P	R	F1-score
SVM	0.6939	0.7603	0.8193	0.8883	0.7602
FCN-8s	0.7886	0.8601	0.9023	0.8593	0.8803
UNet	0.7835	0.8582	0.9125	0.8470	0.8785
DeeplabV3+	0.7982	0.8676	0.8622	0.9112	0.8860
Baseline	0.8054	0.8812	0.9423	0.8426	0.8897
本文方法	0.8165	0.8868	0.9414	0.8602	0.8990