基于高光谱和多头注意力机制的草鱼鲜味强度检测

万仕文; 冯耀泽; 舒国强; 赵名泉; 王益健; 孔丽琴; 朱明

doi:10.13300/j.cnki.hnlkxb.2025.05.028

您当前的位置：

首页 >

文章列表页 >

基于高光谱和多头注意力机制的草鱼鲜味强度检测

食品与农业工程 | 更新时间：2026-11-05

- 基于高光谱和多头注意力机制的草鱼鲜味强度检测
- Detection of umami intensity in grass carp based on hyper-spectrum and multi-attention mechanisms
- 华中农业大学学报 2025年44卷第5期页码：280-287
- 作者机构：
  
  1.华中农业大学信息学院，武汉 430070
  2.华中农业大学工学院，武汉 430070
  3.华中农业大学深圳营养与健康研究院，深圳 518000
  4.中国农业科学院深圳农业基因组研究所/ 岭南现代农业科学与技术广东省实验室深圳分中心，深圳 518120
  5.农业农村部长江中下游农业装备实验室，武汉 430070
  6.农业农村部水产养殖重点实验室，武汉 430070
- 作者简介：
  
  万仕文，E-mail：wsw_2022@webmail.hzau.edu.cn
  冯耀泽，E-mail：Yaoze.feng@mail.hzau.edu.cn
- 基金信息：
  
  湖北省重点研发计划项目(2023BBB038);华中农业大学-深圳营养与健康研究院合作基金项目(SZYJY2021028)
- DOI：10.13300/j.cnki.hnlkxb.2025.05.028
  中图分类号： TP391.4
- 收稿：2024-12-24，
  
  纸质出版：2025-09-15
- 稿件说明：
移动端阅览
万仕文，冯耀泽，舒国强，等.基于高光谱和多头注意力机制的草鱼鲜味强度检测［J］.华中农业大学学报，2025，44（5）：280‐287.

WAN Shiwen,FENG Yaoze,SHU Guoqiang,et al.Detection of umami intensity in grass carp based on hyper-spectrum and multi-attention mechanisms[J].Journal of Huazhong Agricultural University,2025,44(05):280-287.
万仕文，冯耀泽，舒国强，等.基于高光谱和多头注意力机制的草鱼鲜味强度检测［J］.华中农业大学学报，2025，44（5）：280‐287. DOI： 10.13300/j.cnki.hnlkxb.2025.05.028.

WAN Shiwen,FENG Yaoze,SHU Guoqiang,et al.Detection of umami intensity in grass carp based on hyper-spectrum and multi-attention mechanisms[J].Journal of Huazhong Agricultural University,2025,44(05):280-287. DOI： 10.13300/j.cnki.hnlkxb.2025.05.028.

摘要

针对现有鲜味强度检测方法主观性强、耗时长和样本破坏性等问题，使用深度学习和机器学习算法结合高光谱成像技术构建草鱼鲜味强度快速无损检测方法。采集草鱼高光谱数据后，使用竞争性自适应重加权抽样法选取光谱特征波长，开发高斯加权多头注意力网络（gaussian-weighted multi-head attention network，GMANet）并应用支持向量机回归（support vector machine regression，SVR）、偏最小二乘回归（partial least squares regression，PLSR）、随机森林（random forest，RF）、1D-ResNet等传统算法建立和优化草鱼鲜味检测模型。结果显示，GMANet网络的预测均方根误差RMSEP和预测决定系数（

https://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=93869502&type=

https://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=93869518&type=

3.64066648

3.38666677

）分别为0.008 2和0.884 4，优于传统算法中的最优建模方法SVR，其RMSEP和

https://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=93869502&type=

https://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=93869518&type=

3.64066648

3.38666677

分别为0.007 7和0.818 8。

Abstract

Deep learning and machine learning algorithms combined with hyperspectral imaging technology were used to establish a fast and nondestructive method for detecting umami intensity in grass carp to solve the prob

lems of strong subjectivity，long time-consumption and sample destructiveness in the current detection methods of umami intensity.The competitive adaptive reweighted sampling method was used to select the feature wavelengths of spectrum after collecting the hyperspectral data in grass carp.The Gaussian-weighted multi-head attention network （GMANet） was developed.The model of detecting umami intensity in grass carp was established and optimized with the support vector machine regression （SVR），partial least squares regression （PLSR） random forest （RF），1D-ResNet and other algorithms.The results showed that the root mean square error of prediction （

https://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=93869516&type=

https://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=93869533&type=

13.03866673

2.87866688

） and the coefficient of determination of prediction （

https://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=93869517&type=

https://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=93869521&type=

4.14866638

3.97933316

） of GMANet network was 0.008 2 and 0.884 4，better than the RMSEP and

https://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=93869517&type=

https://html.publish.founderss.cn/rc-pub/api/common/picture?pictureId=93869521&type=

4.14866638

3.97933316

of 0.007 7 and 0.818 8 in the optimal modeling method SVR in traditional algorithms.It is indicated that hyperspectral technology has a large application prospect in the detection of umami intensity and the GMANet network can make full use of the spatial image and spectral information of the samples.It will provide a new method for the further application of detection with hyperspectral image.

关键词

Keywords

references

中华人民共和国农业农村部渔业渔政管理局，全国水产技术推广总站，中国水产学会.中国渔业统计年鉴［M］.北京：中国农业出版社，2024.Fishery Administration Bureau of the Ministry of Agriculture and Rural Affairs of the People's Republic of China，National Aquaculture Technology Promotion Station，China Fisheries Society China.Fisheries statistical yearbook ［M］.Beijing： China Agricultural Press，2024 （in Chinese） .

阮秋凤，安玥琦，陈周，等．短时间微流水处理对草鱼鱼肉风味品质的影响［J］.食品科学技术学报，2021，39（3）：30-42.RUAN Q F，AN Y Q，CHEN Z，et al.Effect of short-time micro-flow water treatment on flavor quality of grass carp fish meat［J］.Journal of food science and technology，2021，39（3）：30-42 （in Chinese with English abstract） .

ZHU Y W，WANG J，WANG W L，et al.The evaluation of overall umami intensity in Takifugu obscurus and Ctenopharyngodon idella based on the Steven’s law［J］.Journal of food measurement and characterization，2020，14（1）：527-534.

DONG F J，NIU Y H，BI Y Z，et al.Fusion of spectra and texture features of hyperspectral imaging for quantification and visualization of characteristic amino acid contents in beef［J/OL］.LWT，2024，206：116576［2024-12-24］.https：//doi.org/10.1016/j.lwt.2024.116576.

WANG Y J，WAN S W，PAN S，et al.A quantitative equation for umami evaluation of food and condiments based on the Weber-Fechner law［J/OL］.Food quality and preference，2025，123：105347［2024-12-24］.https：//doi.org/10.1016/j.foodqual.2024.105347.

PHAT C，MOON B，LEE C.Evaluation of umami taste in mushroom extracts by chemical analysis，sensory evaluation，and an electronic tongue system［J］.Food chemistry，2016，192：1068-1077.

丁子予，岳学军，曾凡国，等.基于机器学习和深度学习的玉米种子活力光谱检测［J］.华中农业大学学报，2023，42（3）：230-240.DING Z Y，YUE X J，ZENG F G，et al.Spectral detection of maize seed vigor based on machine learning and deep learning［J］.Journal of Huazhong Agricultural University，2023，42（3）：230-240 （in Chinese with English abstract） .

ZOU Z Y，LI M H，WANG Q L，et al.A non-destructive detection method of protein and TVB-N content changes in refrigerated and frozen-thawed salmon fillets using fluorescence hyperspectral technology［J/OL］.Journal of food composition and analysis，2024，133：106435［2024-12-24］.https：//doi.org/10.1016/j.jfca.2024.106435.

CAO Y M，ZHANG Y，WANG Q，et al.Skin hyperspectral imaging and machine learning to accurately predict the muscular poly-unsaturated fatty acids contents in fish［J/OL］.Current research in food science，2024，9：100929［2024-12-24］.https：//doi.org/10.1016/j.crfs.2024.100929.

詹白勺，章海亮，杨建国.基于高光谱成像技术的三文鱼肉水分含量的可视化研究［J］.光谱学与光谱分析，2017，37（4）：1232-1236.ZHAN B S，ZHANG H L，YANG J G.Visualization of the water content for salmon fish fillets based on hyperspectral imaging technique［J］.Spectroscopy and spectral analysis，2017，37（4）：1232-1236 （in Chinese with English abstract） .

SUN Z B，PAN H D，ZUO M，et al.Non-destructive assessment of equivalent umami concentrations in salmon using hyperspectral imaging technology combined with multivariate algorithms［J/OL］.Spectrochimica acta part A：molecular and biomolecular spectroscopy，2023，285：121890［2024-12-24］.https：//doi.org/10.1016/j.saa.2022.121890.

ZHANG W D，CAO A L，SHI P Y，et al.Rapid evaluation of freshness of largemouth bass under different thawing methods using hyperspectral imaging［J/OL］.Food control，2021，125：108023［2024-12-24］.https：//doi.org/10.1016/j.foodcont.2021.108023.

陈周，胡杨，安玥琦，等.短时间微流水处理对池塘养殖草鱼鱼肉品质的提升作用［J］.水产学报，2020，44（7）：1198-1210.CHEN Z，HU Y，AN Y Q，et al.Quality improvement of short-time micro-flow water treatment on the flesh of grass carp （ Ctenopharyngodon idella ） cultured in a pond［J］.Journal of fisheries of China，2020，44（7）：1198-1210 （in Chinese with English abstract） .

孙宗保，王天真，李君奎，等.高光谱成像的牛肉丸掺假检测［J］.光谱学与光谱分析，2020，40（7）：2208-2214.SUN Z B，WANG T Z，LI J K，et al.Detecting adulterated beef meatball using hyperspectral imaging technology［J］.Spectroscopy and spectral analysis，2020，40（7）：2208-2214 （in Chinese with English abstract） .

孔丽琴，牛晓虎，王程磊，等.高光谱技术在牛肉丸复合掺假类型鉴别中的应用［J］.光谱学与光谱分析，2024，44（8）：2183-2191.KONG L Q，NIU X H，WANG C L，et al.Application of hyperspectral imaging technology in the identification of composite adulteration type in beef balls［J］.Spectroscopy and spectral analysis，2024，44（8）：2183-2191 （in Chinese with English abstract） .

牛晓虎，冯耀泽，鲍雪，等.基于高光谱技术的细菌生物被膜分类检测［J］.华中农业大学学报，2023，42（3）：241-249.NIU X H，FENG Y Z，BAO X，et al.Classification and detection of bacterial biofilms based on hyperspectral fluorescence imaging［J］.Journal of Huazhong Agricultural University，2023，42 （3）：241-249 （in Chinese with English abstract） .

SUN L，ZHAO G R，ZHENG Y H，et al.Spectral-spatial feature tokenization transformer for hyperspectral image classification［J/OL］.IEEE transactions on geoscience and remote sensing，2022，60：3144158［2024-12-24］.https：//doi.org/10.1109/TGRS.2022.3144158.

SHENG H M，LIU X，BAI L B，et al.Small sample state of health estimation based on weighted Gaussian process regression［J/OL］.Journal of energy storage，2021，41：102816［2024-12-24］.https：//doi.org/10.1016/j.est.2021.102816.

VASWANI A.Attention is all you need［DB/OL］.arXiv，2017，1706.03762［2024-12-24］.https：//doi.org/10.48550/arXiv.1706.03762.

黄华成，吴雪梅，张珍，等.基于高光谱技术的绿色辣椒识别研究［J］.计算机与数字工程，2024，52（8）：2473-2478.HUANG H C，WU X M，ZHANG Z，et al.Research on recognition of green pepper based on hyperspectral technology［J］.Computer & digital engineering，2024，52（8）：2473-2478 （in Chinese with English abstract） .

黄华丹，邓健豪，曹庸，等.基于近红外光谱技术快速检测广式酱油发酵过程中主要理化指标的含量［J］.现代食品科技，2024，40（12）：386-401.HUANG H D，DENG J H，CAO Y，et al.Rapid determination of main physicochemical indexes of Cantonese soy sauce in the fermentation process based on near-infrared spectroscopy［J］.Modern food science and technology，2024，40（12）：386-401 （in Chinese with English abstract） .

LI Z Y，WANG D Y，ZHU T T，et al.Review of deep learning-based methods for non-destructive evaluation of agricultural products［J］.Biosystems engineering，2024，245：56-83.

YU Z J，XIE T Y，ZHU Q B，et al.Aquatic plants detection in crab ponds using UAV hyperspectral imagery combined with transformer-based semantic segmentation model［J/OL］.Computers and electronics in agriculture，2024，227：109656［2024-12-24］.https：//doi.org/10.1016/j.compag.2024.109656.

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于改进YOLOv8s的马铃薯种薯薯块芽眼检测

基于改进YOLO11的番茄叶片病害检测方法

EMD-YOLO：基于YOLOv8n改进的轻量化作物叶部病害检测算法

基于双重注意力机制的鱼类体表病理识别方法

不同杀菌剂对百合炭腐病菌的毒力和防效评价