| 30 | 0 | 36 |
| 下载次数 | 被引频次 | 阅读次数 |
粗梗水蕨（Ceratopteris thalictroides）作为典型的汉江流域湿地生态指示植物，具有重要的生态功能和保护价值。然而，其叶片结构复杂，姿态变化显著，为图像分割带来了显著挑战。为提升模型对粗梗水蕨图像的分割能力，提出了一种基于改进注意力机制的UNet3+图像分割方法。该方法在原始SCSA(spatial and channel self-attention)模块的基础上引入非对称卷积结构，增强模型对植物主体结构和边缘细节的建模能力，并将该注意力嵌入UNet3+的下采样阶段，以强化编码过程中关键区域的特征表达。构建了包含不同视角、不同环境下的粗梗水蕨图像数据集，采用Dice、IoU、F1score和mIoU等指标对模型性能进行评估。实验结果显示，所提出的SCSAA-UNet3+模型在多个评价指标上均优于传统分割方法，验证了其在粗梗水蕨图像分割任务中的有效性。
Abstract：Ceratopteris thalictroides(C. thalictroides),a representative wetland ecological indicator species in the Hanjiang River Basin, plays an important role in ecosystem functioning and conservation. However,its leaf structure is complex and its posture varies significantly,which poses considerable challenges for image segmentation. To address these issues,this study proposes an improved UNet3+ segmentation framework incorporating an enhanced attention mechanism. On the basis of the original spatial and channel selfattention(SCSA)module,an asymmetric convolution structure is introduced to enhance the model′s ability to capture the main plant structure and edge details. This modified attention module is embedded into the downsampling stages of UNet3+ to strengthen the feature representation of key regions during encoding. A dedicated dataset of C. thalictroides images captured under multiple viewpoints and environmental conditions was constructed,and the model′s performance was evaluated using Dice,IoU,F1 score,and mIoU. Experimental results show that the proposed SCSAA-UNet3+ model achieves superior segmentation performance compared with conventional methods,demonstrating its effectiveness in the image segmentation task of C. thalictroides.
[1]国家林业和草原局，农业农村部.国家重点保护野生植物（2021版）[EB/OL].(2021-9-9)[2024-10-13]. https：∥www.gov.cn/zhengce/zhengceku/2021-09/09/content＿5636409.htm.
[2]白国梁，许伟翔，张聪，等.水生植物修复受污染水体的应用研究[J].农业现代化研究，2025,46(2):351-365.
[3]Rodrigo M A. Wetland restoration with hydrophytes:A review[J]. Plants,2021,10(6):1035.
[4]Ronneberger O,Fischer P,Brox T. U-Net:Convolutional networks for biomedical image segmentation[C]∥Navab N,Hornegger J,Wells W M,et al. Medical image computing and computer-assisted interventionMICCAI 2015. Cham:Springer,2015:234-241.
[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.
[6]Zhou Z,Siddiquee M M R,Tajbakhsh N,et al. UNet++:A nested U-Net architecture for medical image segmentation[C]∥Stoyanov D,Syeda-Mahmood T,Montesinos P,et al. Deep learning in medical image analysis and multimoal learning for clinical decision support. Cham:Springer,2018:3-11.
[7]Liu Z,Lin Y,Cao Y,et al. Swin transformer:Hierarchical vision transformer using shifted windows[C]∥Proceedings of the IEEE/CVF international conference on computer vision,2021:10012-10022.
[8]Gao H,Li R,Shen Q,et al. Deep-learning-based automatic extraction of aquatic vegetation from Sentinel-2images:A case study of Lake Honghu[J]. Remote Sensing,2024,16(5):867.
[9]Muhadi N A,Abdullah A F,Bejo S K,et al. Deep learning semantic segmentation for water level estimation using surveillance camera[J]. Applied Sciences,2021,11(20):9691.
[10]Yang T,Zhou S,Xu A,et al. An approach for plant leaf image segmentation based on YOLOv8 and the improved DeepLabV3+[J]. Plants,2023,12(20):3438.
[11]Lu J W,Lu B B,Ma W L,et al. EAIS-Former:An efficient and accurate image segmentation method for fruit leaf diseases[J]. Computers and Electronics in Agriculture,2024,218:108739.
[12]Yi X,Wang J,Wu P,et al. AC-UNet:an improved UNet-based method for stem and leaf segmentation in Betula luminifera[J]. Frontiers in Plant Science,2023,14:1268098.
[13]Zhou H,Yang J,Zhang T,et al. EAS-CNN:automatic design of convolutional neural network for remote sensing images semantic segmentation[J]. International Journal of Remote Sensing,2023,44(13):3911-3938.
[14]Huang H,Lin L,Tong R,et al. UNet3+:A full-scale connected UNet for medical image segmentation[C]∥ICASS P 2020-2020 IEEE International conference on acoustics,speech and signal processing. Barcelona:IEEE,2020:1055-1059.
[15]Si Y,Xu H,Zhu X,et al. SCSA:Exploring the synergistic effects between spatial and channel attention[J].Neurocomputing,arXiv preprint,arXiv:2407.05128,2024.
[16]Oktay O,Schlemper J,Folgoc L L,et al. Attention U-Net:Learning where to look for the pancreas[J].arxiv preprint arxiv:1804.03999,2018.
[17]Yu H,Dong M,Zhao R,et al. Research on precise phenotype identification and growth prediction of lettuce based on deep learning[J]. Environmental Research,2024,252:118845.
[18]Chen J,Lu Y,Yu Q,et al. TransUNet:Transformers make strong encoders for medical image segmentation[J]. arXiv preprint,arXiv:2102.04306,2021.
基本信息:
DOI：10.16389/j.cnki.cn42-1737/n.2026.04.002
中图分类号:Q948.8;X173
引用信息:
[1]胡慧莉,叶曦,董元火,等.基于改进注意力机制的粗梗水蕨图像分割方法[J].江汉大学学报(自然科学版),2026,54(04):15-25.DOI:10.16389/j.cnki.cn42-1737/n.2026.04.002.
基金信息:
湖北省自然科学基金项目（2023AFB462）; 江汉大学研究生科研创新基金项目（KYCXJJ202434）
2026-07-02
2026-07-02
2026-07-02