[1]蒋雄杰,张辉,刘立柱,等.超像素稀疏注意力引导的中药高光谱图像分割方法[J].智能系统学报,2026,21(1):272-283.[doi:10.11992/tis.202507028]
JIANG Xiongjie,ZHANG Hui,LIU Lizhu,et al.Superpixel sparse attention-guided hyperspectral image segmentation network for traditional Chinese medicine[J].CAAI Transactions on Intelligent Systems,2026,21(1):272-283.[doi:10.11992/tis.202507028]
点击复制
JIANG Xiongjie,ZHANG Hui,LIU Lizhu,et al.Superpixel sparse attention-guided hyperspectral image segmentation network for traditional Chinese medicine[J].CAAI Transactions on Intelligent Systems,2026,21(1):272-283.[doi:10.11992/tis.202507028]
超像素稀疏注意力引导的中药高光谱图像分割方法
《智能系统学报》[ISSN 1673-4785/CN 23-1538/TP] 卷:
21
期数:
2026年第1期
页码:
272-283
栏目:
人工智能院长论坛
出版日期:
2026-03-05
- Title:
- Superpixel sparse attention-guided hyperspectral image segmentation network for traditional Chinese medicine
- Keywords:
- traditional Chinese medicine quality identification; hyperspectral images; attention mechanism; semantic segmentation; feature extraction; cross-scale fusion; superpixel; sparse attention
- 分类号:
- TP391.4；TH744
- 摘要:
- 在中医药质量检测中，针对传统红绿蓝三通道图像(red green blue，RGB)检测方法因缺乏光谱信息难以实现精准鉴别、高光谱方法在“类间差异小、类内差异大”场景下面临精度与效率不足的问题，本文构建了线扫描式高光谱成像系统，建立5个具有像素级标注的中药材高光谱数据集，并提出了一种空间–光谱超像素稀疏注意力引导的中药高光谱图像分割网络。该网络通过三阶段架构实现中药质量无损高精度检测，基于空间距离和光谱距离将高光谱图像的同质区域分割为超像素块，增强语义一致性，提升特征学习效率；利用双流超像素稀疏注意力模块，通过构建超像素块间关联矩阵过滤不相关区域，实现强关联区域间的空间维度全局特征提取与光谱维度的波段显著性建模；通过多尺度特征融合解码器实现像素级检测。实验结果表明，在真伪鉴别、产地溯源和炮制品鉴别数据集上，本文方法平均像素精度(mean pixel accuracy，MPA)和平均交互比(mean intersection over union，MIoU)分别为0.968、0.933，均优于现有方法。本文方法可为中药质量精准检测提供参考。
- Abstract:
- In the context of quality control in traditional Chinese medicine, to address the limitations of conventional RGB-based methods—which struggle to achieve accurate identification due to the lack of spectral information—and the challenges faced by existing hyperspectral approaches in scenarios characterized by small inter-class differences and large intra-class variations, a line-scan hyperspectral imaging system was developed and five pixel-level annotated hyperspectral datasets were constructed for TCM materials. Moreover, a spatial–spectral superpixel, sparse attention-guided hyperspectral image segmentation network was proposed for non-destructive, high-precision quality detection via a three-stage architecture. First, the network segments homogeneous regions of the hyperspectral image into superpixel blocks based on the spatial and spectral distances to enhance the semantic consistency and improve feature learning. Thereafter, a dual-stream superpixel sparse attention module constructs a superpixel correlation matrix, filters out irrelevant regions, and enables the extraction of global spatial features and the modeling of spectral band saliency in strongly correlated areas. Finally, a multi-scale feature fusion decoder achieves pixel-level segmentation. As demonstrated experimentally, the proposed method achieves an mean pixel accuracy(MPA) of 0.968 and an mean intersection over union(MIoU) of 0.933 across datasets for authenticity identification, origin tracing, and processing-type classification, outperforming existing approaches. These results indicate that the proposed framework offers a robust and effective solution for precise quality assessment of traditional Chinese medicinal materials.
- 参考文献/References:
-
[1] LI Yaolei, FAN Jing, CHENG Xianlong, et al. New revolution for quality control of TCM in industry 4.0: focus on artificial intelligence and bioinformatics[J]. Trends in analytical chemistry, 2024, 181: 118023
[2] YU Yang, YAO Changliang, GUO Dean. Insight into chemical basis of traditional Chinese medicine based on the state-of-the-art techniques of liquid chromatography-mass spectrometry[J]. Acta pharmaceutica sinica B, 2021, 11(6): 1469-1492
[3] LI Tao, ZHUANG Shuaixing, WANG Yiwei, et al. Flavonoid profiling of a traditional Chinese medicine formula of Huangqin Tang using high performance liquid chromatography[J]. Acta pharmaceutica sinica B, 2016, 6(2): 148-157
[4] PRIDEAUX B, LENAERTS A, DARTOIS V. Imaging and spatially resolved quantification of drug distribution in tissues by mass spectrometry[J]. Current opinion in chemical biology, 2018, 44: 93-100
[5] WANG Tianlong, LI Yicong, LIN Chunsheng, et al. Comprehensive analysis of natural polysaccharides from TCMs: a generic approach based on UPLC-MS/MS[J]. Carbohydrate polymers, 2022, 277: 118877
[6] ZHANG Fukai, JIN Xiaobo, LIN Gang, et al. Hybrid attention network for citrus disease identification[J]. Computers and electronics in agriculture, 2024, 220: 108907
[7] 黄鸿, 唐玉枭, 段宇乐. 面向高光谱分类的局部几何稀疏保持嵌入[J]. 自动化学报, 2022, 48(10): 2496-2507 HUANG Hong, TANG Yuxiao, DUANG Yule. Local geometry and sparsity preserving embedding for hyperspectral image classification[J]. Acta automatica sinica, 2022, 48(10): 2496-2507
[8] HU Wei, HUANG Yangyu, WEI Li, et al. Deep convolutional neural networks for hyperspectral image classification[J]. Journal of sensors, 2015, 2015: 258619
[9] SLAVKOVIKJ V, VERSTOCKT S, DE NEVE W, et al. Hyperspectral image classification with convolutional neural networks[C]//Proceedings of the 23rd ACM International Conference on Multimedia. Brisbane: ACM, 2015.
[10] HE Mingyi, LI Bo, CHEN Huahui. Multi-scale 3D deep convolutional neural network for hyperspectral image classification[C]//2017 IEEE International Conference on Image Processing. Beijing: IEEE, 2017.
[11] LIAO Diling, SHI Cuiping, WANG Liguo. A spectral–spatial fusion transformer network for hyperspectral image classification[J]. IEEE transactions on geoscience and remote sensing, 2023, 61: 5515216
[12] 王瑞婷, 王海燕, 陈晓, 等. 基于混合卷积与三重注意力的高光谱图像分类网络[J]. 智能系统学报, 2023, 18(2): 260-269 WANG Ruiting, WANG Haiyan, CHEN Xiao, et al. Hyperspectral image classification based on hybrid convolutional neural network with triplet attention[J]. CAAI transactions on intelligent systems, 2023, 18(2): 260-269
[13] 涂潮, 刘万军, 赵琳琳, 等. 有限训练样本下的多尺度空洞密集网络高光谱影像分类[J]. 仪器仪表学报, 2024, 45(4): 206-216 TU Cao, LIU Wangjun, ZHAO Linlin, et al. Multiscale dilated dense network for hyperspectral image classification[J]. Chinese journal of scientific instrument, 2024, 45(4): 206-216
[14] ROY S K, KRISHNA G, DUBEY S R, et al. HybridSN: exploring 3-D–2-D CNN feature hierarchy for hyperspectral image classification[J]. IEEE geoscience and remote sensing letters, 2020, 17(2): 277-281
[15] HONG Danfeng, HAN Zhu, YAO Jing, et al. SpectralFormer: rethinking hyperspectral image classification with transformers[J]. IEEE transactions on geoscience and remote sensing, 2021, 60: 5518615
[16] RONNEBERGER O, FISCHER P, BROX T. U-Net: convolutional networks for biomedical image segmentation[C]//International Conference on Medical Image Computing and Computer Assisted Intervention. Munich: MICCAI Society, 2015.
[17] WANG Qian, SUN Li, WANG Yan, et al. Identification of melanoma from hyperspectral pathology image using 3D convolutional networks[J]. IEEE transactions on medical imaging, 2021, 40(1): 218-227
[18] YUN Boxiang, LEI Baiying, CHEN Jieneng, et al. SpecTr: spectral transformer for microscopic hyperspectral pathology image segmentation[J]. IEEE transactions on circuits and systems for video technology, 2024, 34(6): 4610-4624
[19] SELCI S. The future of hyperspectral imaging[J]. Journal of imaging, 2019, 5(11): 84
[20] DELWICHE S R, KIM M S, DONG Yanhong. Fusarium damage assessment in wheat kernels by Vis/NIR hyperspectral imaging[J]. Sensing and instrumentation for food quality and safety, 2011, 5(2): 63-71
[21] ZHANG Jun, XU Binbo, WANG Zhiying, et al. Application of hyperspectral imaging in the detection of aflatoxin B1 on corn seed[J]. Journal of food measurement and characterization, 2022, 16(1): 448-460
[22] 史卓林, 杨增玲, 任朝霞, 等. 推扫式双相机高光谱成像系统设计与试验[J]. 农业机械学报, 2024, 55(S1): 288-294, 305 SI Zhuoling, YANG Zengling, RENG Zhaoxia, et al. Design and test of push-broom dual-camera hyperspectral imaging system[J]. Transactions of the Chinese society for agricultural machinery, 2024, 55(S1): 288-294, 305
[23] VINCENT L, SOILLE P. Watersheds in digital spaces: an efficient algorithm based on immersion simulations[J]. IEEE transactions on pattern analysis and machine intelligence, 1991, 13(6): 583-598
[24] COMANICIU D, MEER P. Mean shift: a robust approach toward feature space analysis[J]. IEEE transactions on pattern analysis and machine intelligence, 2002, 24(5): 603-619
[25] VEDALDI A, SOATTO S. Quick shift and kernel methods for mode seeking[C]//10th European Conference on Computer Vision. Berlin: ECVA, 2008.
[26] LEVINSHTEIN A, STERE A, KUTULAKOS K N, et al. TurboPixels: fast superpixels using geometric flows[J]. IEEE transactions on pattern analysis and machine intelligence, 2009, 31(12): 2290-2297
[27] ACHANTA R, SHAJI A, SMITH K, et al. SLIC superpixels compared to state-of-the-art superpixel methods[J]. IEEE transactions on pattern analysis and machine intelligence, 2012, 34(11): 2274-2282
[28] HU Jie, SHEN Li, SUN Gang. Squeeze-and-excitation networks[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Salt Lake City: IEEE, 2018.
[29] WOO S, PARK J, LEE J Y, et al. CBAM: convolutional block attention module[C]//European Conferenceon Computer Vision. Munich: ECVA, 2018.
[30] DOSOVITSKIY Alexey, BEYER Lucas, KOLESNIKOV Alexander, et al. An image is worth 16x16 words: transformers for image recognition at scale[EB/OL]. (2020-10-20)[2025-06-16]. https://arxiv.org/abs/2010.11929arXiv.
[31] XIE Enze, WANG Wenhai, YU Zhiding, et al. SegFormer: simple and efficient design for semantic segmentation with transformers[J]. Advances in neural information processing systems, 2021, 34: 12077-12090
[32] ZHENG Sixiao, LU Jiachen, ZHAO Hengshuang, et al. Rethinking semantic segmentation from a sequence-to-sequence perspective with transformers[C]//2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Nashville: IEEE, 2021.
[33] CHENG Bowen, MISRA I, SCHWING A G, et al. Masked-attention mask transformer for universal image segmentation[C]//2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New Orleans: IEEE, 2022.
[34] LUONG Minh-thang, PHAM H, MANNING C D. Effective approaches to attention-based neural machine translation[EB/OL]. (2015-09-20)[2025-05-25]. https://arxiv.org/abs/1508.04025.
[35] JIAO Jiayu, TANG Yuming, LIN Kunyu, et al. DilateFormer: multi-scale dilated transformer for visual recognition[J]. IEEE transactions on multimedia, 2023, 25: 8906-8919
[36] XIA Zhuofan, PAN Xuran, SONG Shiji, et al. Vision transformer with deformable attention[C]//2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New Orleans: IEEE, 2022.
- 相似文献/References:
-
[1]姚伏天,钱沄涛.高斯过程及其在高光谱图像分类中的应用[J].智能系统学报,2011,6(5):396.
YAO Futian,QIAN Yuntao.Gaussian process and its applications in hyperspectral image classification[J].CAAI Transactions on Intelligent Systems,2011,6():396.
[2]赵春晖,张燚.相关向量机分类方法的研究进展与分析[J].智能系统学报,2012,7(4):294.
ZHAO Chunhui,ZHANG Yi.Research progress and analysis on methods for classification of RVM[J].CAAI Transactions on Intelligent Systems,2012,7():294.
[3]陈姗姗,宁纪锋,彭艺伟,等.基于近红外高光谱图像的苹果轻微损伤检测[J].智能系统学报,2013,8(4):356.[doi:10.3969/j.issn.1673-4785.201304041]
CHEN Shanshan,NING Jifeng,PENG Yiwei,et al.Detection of slight bruises on apples using near-infrared hyperspectral image[J].CAAI Transactions on Intelligent Systems,2013,8():356.[doi:10.3969/j.issn.1673-4785.201304041]
备注/Memo
收稿日期:2025-7-25。
基金项目:国家自然科学基金项目(62027810)；国家自然科学基金重大研究计划项目(92148204)；国家自然科学基金重点项目(62433010).
作者简介:蒋雄杰，硕士研究生，主要研究方向为高光谱图像处理、计算机视觉。E-mail：1036046572@qq.com。;张辉，教授，博士生导师，湖南大学人工智能与机器人学院院长，入选国家高层次人才计划，主要研究方向为机器视觉、高光谱、图像处理和机器人视觉。主持科技创新2030—“新一代人工智能”重大项目课题、国家自然科学基金重点项目、国家重点研发计划子课题等20余项，发表学术论文 60 余篇。E-mail：zhanghui1983@hnu.edu.cn。;王耀南，教授，博士生导师，中国工程院院士。长期从事机器人感知与控制技术教学科研工作，主持完成国家重大科技项目20余项，成果获国家技术发明二等奖1项、国家科技进步二等奖5项、何梁何利基金科学与技术进步奖、国际IEEE机器人与自动化领域“工业应用最高奖”，省部级一等奖12项。发表学术论文500余篇。E-mail：yaonan@hnu.edu.cn。
通讯作者:张辉. E-mail：zhanghui1983@hnu.edu.cn
更新日期/Last Update:
2026-01-05