基于自蒸馏边缘细化的遥感图像语义分割

doi:10.16088/j.issn.1001-6600.2025040201

摘要/Abstract

摘要： 针对遥感图像语义分割过程中产生的边缘特征丢失与大量参数冗余现象,本文提出一种基于自蒸馏边缘细化的分割方法。首先基于EfficientNetB4构建主干网络;然后在自教师网络分支中引入轻量级边缘精细化模块(edge refinement module, ERM)以捕捉中间特征图的局部信息,保留被浅层神经网络过滤的中间边缘信息,从而提高遥感图像边缘像素分割精度;最后,使用每幅图像的二值类别标签为预测矩阵创建自适应多视角(self-adaptive multi-view, SAMV)向量,作为一种新知识指导编码器网络的训练,能更好地描述类内与类间分布,拟合层间与层内关系。在公开数据集DeepGlobe与Vaihingen上平均交并比分别达到72.4%和83.3%,对比实验表明,本文提出的方法能增强边缘特征的同时兼顾分割精度、模型参数与推理速度,在轻量化模型的同时具有良好的特征提取能力。

关键词: 自蒸馏, 边缘细化, 遥感图像, 语义分割, 自适应多视角

Abstract: A segmentation method based on self-distillation edge refinement is proposed in this paper to tackle the challenges of edge feature loss and excessive parameter redundancy encountered during semantic segmentation of remote sensing images. Firstly, a backbone network is constructed using EfficientNetB4 as the foundation. Subsequently, a lightweight edge refinement module is integrated into the self-teacher network branch. This module is designed to capture local information from intermediate feature maps while retaining the intermediate edge details filtered by shallow neural networks, with the purpose to improve the accuracy of edge pixel segmentation in remote sensing images. Finally, an adaptive multi-view vector is created to serve as a novel knowledge guide for encoder network training. This is achieves by utilizing the binary category labels of each image as the prediction matrix. The adaptive multi-view vector provides a better description of intra-class and inter-class distributions, as well as fitting inter-layer and intra-layer relationships. On the public datasets DeepGlobe and Vaihingen, the proposed method achieves an average intersection ratio of 72.4% and 83.3%, respectively. Comparative experiments demonstrate that the method introduced in this study enhances edge features while maintaining a balance among segmentation accuracy, model parameters, and inference speed. It has good feature extraction ability while lightweighting the model.

Key words: self-distillation, edge refinement, remote sensing images, semantic segmentation, adaptive multi-view

中图分类号: TP751;TP391

宋冠武, 李建军. 基于自蒸馏边缘细化的遥感图像语义分割[J]. 广西师范大学学报（自然科学版）, 2026, 44(2): 65-76.

SONG Guanwu, LI Jianjun. Semantic Segmentation of Remote Sensing Images Basedon Self-distillation Edge Refinement[J]. Journal of Guangxi Normal University(Natural Science Edition), 2026, 44(2): 65-76.

参考文献

[1] 苏林茂, 曹经.人工智能在遥感图像分类中的研究[J].中国科技信息, 2022(12):98-102.DOI:10.3969/j.issn.1001-8972.2022.12.031.
[2] 刘金香, 班伟, 陈宇, 等.融合多维度CNN的高光谱遥感图像分类算法[J].中国激光, 2021, 48(16):1610003.DOI:10.3788/CJL202148.1610003.
[3] 张淳民, 穆廷魁, 颜廷昱, 等.高光谱遥感技术发展与展望[J].航天返回与遥感, 2018, 39(3):104-114.DOI:10.3969/j.issn.1009-8518.2018.03.012.
[4] 龚健雅.人工智能时代测绘遥感技术的发展机遇与挑战[J].武汉大学学报(信息科学版), 2018, 43(12):1788-1796.DOI:10.13203/j.whugis20180082.
[5] CHEN L C, PAPANDREOU G, KOKKINOS I, et al.Deeplab:semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected CRFs[J].IEEE Transactions on Pattern Analysis and Machine Intelligence, 2018, 40(4):834-848.DOI:10.1109/TPAMI.2017.2699184.
[6] YANG M K, YU K, ZHANG C, et al.DenseASPP for semantic segmentation in street scenes[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Los Alamitos, CA:IEEE Computer Society, 2018:3684-3692.DOI:10.1109/CVPR.2018.00388.
[7] XIAO X Y, ZHAO Y Q, ZHANG F, et al.BASeg:boundary aware semantic segmentation for autonomous driving[J].Neural Networks, 2023, 157:460-470.DOI:10.1016/j.neunet.2022.10.034.
[8] GAO G W, XU G A, LI J C, et al.FBSNet:a fast bilateral symmetrical network for real-time semantic segmentation[J].IEEE Transactions on Multimedia, 2023, 25:3273-3283.DOI:10.1109/TMM.2022.3157995.
[9] SPASEV V, DIMITROVSKI I, CHORBEV I, et al.Semantic segmentation of unmanned aerial vehicle remote sensing images using SegFormer[C]//Intelligent Systems and Pattern Recognition.Cham:Springer Nature Switzerland AG, 2025:108-122.DOI:10.1007/978-3-031-82156-1_9.
[10] HINTON G, VINYALS O, DEAN J.Distilling the knowledge in a neural network[EB/OL].(2015-03-09)[2025-04-02].https://arxiv.org/abs/1503.02531.DOI:10.48550/arXiv.1503.02531.
[11] YANG C G, ZHOU H L, AN Z L, et al.Cross-image relational knowledge distillation for semantic segmentation[C]//2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).Los Alamitos, CA:IEEE Computer Society, 2022:12309-12318.DOI:10.1109/CVPR52688.2022.01200.
[12] JI M I, SHIN S, HWANG S, et al.Refine myself by teaching myself:feature refinement via self-knowledge distillation[C]//2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).Los Alamitos, CA:IEEE Computer Society, 2021:10659-10668.DOI:10.1109/CVPR46437.2021.01052.
[13] 孙显, 杨竹君, 李俊希, 等.基于知识自蒸馏的轻量化复杂遥感图像精细分类方法[J].指挥与控制学报, 2021, 7(4):365-373.DOI:10.3969/j.issn.2096-0204.2021.04.0365.
[14] ZHAO Q, MA Y J, LYU S C, et al.Embedded self-distillation in compact multibranch ensemble network for remote sensing scene classification[J].IEEE Transactions on Geoscience and Remote Sensing, 2022, 60:4506415.DOI:10.1109/TGRS.2021.3126770.
[15] YANG C G, AN Z L, ZHOU H L, et al.MixSKD:Self-knowledge distillation from mixup for image recognition[C]//Computer Vision - ECCV 2022:LNCS Volume 13684.Cham:Springer Nature Switzerland AG, 2022:534-551.DOI:10.1007/978-3-031-20053-3_31.
[16] RASHEEDHA A, REETHIKA A, RAJESHWARI R, et al.A hierarchical aggregation network with edge guidance for sturdy medical image segmentation[C]//2024 2nd International Conference on Advances in Computation, Communication and Information Technology (ICAICCIT).Piscataway, NJ:IEEE, 2024:1105-1109.DOI:10.1109/ICAICCIT64383.2024.10912254.
[17] ROH W, JUNG H, NAM G, et al.Edge-aware 3D instance segmentation network with intelligent semantic prior[C]//2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).Los Alamitos, CA:IEEE Computer Society, 2024:20644-20653.DOI:10.1109/CVPR52733.2024.01951.
[18] 张德雄.基于语义分割的森林场景区域识别技术研究[D].南京:东南大学, 2021.DOI:10.27014/d.cnki.gdnau.2021.003164.
[19] 张博洋.基于卷积神经网络的林木图像语义分割研究[D].哈尔滨:东北林业大学, 2022.DOI:10.27009/d.cnki.gdblu.2022.001505.
[20] HUYNH C, TRAN A T, LUU K, et al.Progressive semantic segmentation[C]//2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).Los Alamitos, CA:IEEE Computer Society, 2021:16750-16759.DOI:10.1109/CVPR46437.2021.01648.
[21] LIU W X, LI Q, LIN X D, et al.Ultra-high resolution image segmentation via locality-aware context fusion and alternating local enhancement[J].International Journal of Computer Vision, 2024, 132(11):5030-5047.DOI:10.1007/s11263-024-02045-3.
[22] YANG L, BAI Y W, REN F L, et al.LCFNets:compensation strategy for real-time semantic segmentation of autonomous driving[J].IEEE Transactions on Intelligent Vehicles, 2024, 9(4):4715-4729.DOI:10.1109/TIV.2024.3363830.
[23] KARINE A, NAPOLÉON T, JRIDI M.Channel-spatial knowledge distillation for efficient semantic segmentation[J].Pattern Recognition Letters, 2024, 180:48-54.DOI:10.1016/j.patrec.2024.02.027.
[24] WU H L, HUANG P, ZHANG M, et al.CMTFNet:CNN and multiscale transformer fusion network for remote-sensing image semantic segmentation[J].IEEE Transactions on Geoscience and Remote Sensing, 2023, 61:2004612.DOI:10.1109/TGRS.2023.3314641.
[25] CHEN B Z, LIU Y S, ZHANG Z, et al.TransAttUnet:multi-level attention-guided U-Net with transformer for medical image segmentation[J].IEEE Transactions on Emerging Topics in Computational Intelligence, 2024, 8(1):55-68.DOI:10.1109/TETCI.2023.3309626.
[26] WU H S, ZHAO Z B, WANG Z Z.META-Unet:multi-scale efficient transformer attention Unet for fast and high-accuracy polyp segmentation[J].IEEE Transactions on Automation Science and Engineering, 2024, 21(3):4117-4128.DOI:10.1109/TASE.2023.3292373.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed