图像预处理整合策略结合改进YOLOv8模型用于微藻种类识别

宁静; 钟月妍; 刘学英; 谢丽霞; 王童

doi:null

您当前的位置：

首页 >

文章列表页 >

图像预处理整合策略结合改进YOLOv8模型用于微藻种类识别

研究报告 | 更新时间：2025-04-28

- 图像预处理整合策略结合改进YOLOv8模型用于微藻种类识别
- Integration Strategy Combined with Improved YOLOv8 Model for Identification of Microalgae Species
- 分析测试学报 2025年
- 作者机构：
  
  1.湖南工业大学生命科学与化学学院，百合种质资源创新与深加工湖南省工程研究中心，湖南株洲 412007
  2.湖南大学化学化工学院，化学生物传感与计量学国家重点实验室，湖南长沙 410082
- 作者简介：
  
  谢丽霞，博士，湖南工业大学副教授，主要研究方向为化学计量学结合色谱、光谱新技术新方法及在食品、环境、生物分析中的应用；碳量子点荧光传感器；图像识别，数据处理。E-mail：m15116269721@163.com
- 基金信息：
  
  湖南省生态环境厅科研项目(HBKT-2021010)
- DOI：
  中图分类号：
- 稿件说明：
移动端阅览
宁静, 钟月妍, 刘学英, 谢丽霞, 王童. 图像预处理整合策略结合改进YOLOv8模型用于微藻种类识别[J/OL]. 分析测试学报, 2025.

NING Jing, ZHONG Yueyan, LIU Xueying, XIE Lixia, WANG Tong. Integration Strategy Combined with Improved YOLOv8 Model for Identification of Microalgae Species[J/OL]. Journal of instrumental analysis, 2025.
宁静, 钟月妍, 刘学英, 谢丽霞, 王童. 图像预处理整合策略结合改进YOLOv8模型用于微藻种类识别[J/OL]. 分析测试学报, 2025. DOI：

NING Jing, ZHONG Yueyan, LIU Xueying, XIE Lixia, WANG Tong. Integration Strategy Combined with Improved YOLOv8 Model for Identification of Microalgae Species[J/OL]. Journal of instrumental analysis, 2025. DOI：

摘要

为了解决传统的微藻检测方法依赖于人工镜检，存在分析时间长且检测结果易受检测人员技术经验影响等问题，提出了一种图像预处理整合策略结合改进YOLOv8模型的深度学习方法用于微藻识别。采用高斯模糊、拉普拉斯算子和主成分分析多方法整合策略对微藻显微图像进行预处理。在改进模型中，引入SPD-Conv模块减少细粒度信息的丢失以提高低分率图像和小尺寸微藻的检测性能，采用Slim-neck结构减少参数数量和计模型体积，同时加入SimSPPF加速模型收敛，提高运行效率。实验结果表明，多方法整合的预处理策略能够显著减少图像中的噪声，同时增强微藻轮廓清晰度。在相同条件下，改进的YOLOv8模型的平均精度均值（

mAP

）达到92.2%，检测效率比原始YOLOv8模型提高了5.1%，且对于小尺寸微藻表现出更优的检测性能。相较于Faster-RCNN、SSD、RTDETR-l、YOLOv3、YOLOv5、YOLOv6和YOLOv7模型，改进YOLOv8模型的

mAP

分别提升了40.2%、6.8%、14.5%、1.2%、5.7%、4.7%和0.8%。该方法为开发微藻种类检测技术提供了有价值的参考。

Abstract

To address the limitations of traditional microalgae detection methods

which rely on manual microscopy

result in prolonged analysis times

and produce results that are highly susceptible to the technical expertise of personnel

an integrated image preprocessing strategy combined with an enhanced YOLOv8 deep learning model for microalgae identification was proposed. A multi-method integration strategy of Gaussian fuzzy

Laplacian operator and principal component analysis was used to preprocess microalgae images. In the improved model

the SPD-Conv module was incorporated to mitigate the loss of fine-grained information

thereby improving the detection performance for low-resolution images and small-sized microalgae. A slim-neck architecture was employed to reduce the parameter count and model size

while the SimSPPF was introduced to expedite model convergence and enhance operational efficiency. The experimental results demonstrated that the multi-method integrated preprocessing strategy was able to substantially reduce image noise and enhance the definition of microalgal contours. Under identical conditions

the improved YOLOv8 model achieved a mean average precision (

mAP

) of 92.2%

representing a 5.1% improvement over the original YOLOv8

model. Especially

it demonstrated superior performance in detecting small-sized microalgae. In comparison to Faster-RCNN

SSD

RTDETR-l

YOLOv3

YOLOv5

YOLOv6

and YOLOv7 models

the

mAP

of improved YOLOv8 model increased by 40.2%

6.8%

14.5%

1.2%

5.7%

4.7%

and 0.8%

respectively. This method offers valuable insights for advancing microalgae species detection technology.

关键词

Keywords

references

Lobus N V , Kulikovskiy M S . Biology , 2023 , 12 ( 1 ): 92 .

Chen N S ， Zhang M J ， Liu S Y ， Cui Z M . Oceanol. Limnol. Sin （陈楠生，张梦佳，刘淑雅，崔宗梅. 海洋与湖沼）， 2023 , 54 ： 599 - 624 .

Yan T , Li X D , Tan Z J , Yu R C , Zou J Z . Harmful Algae , 2022 , 111 : 102148 .

Smith V H , Schindler D W . Trends Ecol. Evol. , 2009 , 24 ( 4 ): 201 - 207 .

Barsanti L , Birindelli L , Gualtieri P . Environ. Sci. Processes Impacts , 2021 , 23 ( 10 ): 1443 - 1457 .

Chong J W R , Khoo K S , Chew Kit W , Vo D N , Balakrishnan D , Banat F , Munawaroh H S H , Iwamoto K , Show P L . Bioresour. Technol. , 2023 , 369 : 128418 .

Xu Z , Jiang Y , Ji J , Forsberg E , Li Y , He S . Opt. Express , 2020 , 28 ( 21 ): 30686 - 30700 .

Wu Z G . Research and Application of Microalgae Lightweight Detection Algorithm Based on Improved YOLO v7 . Shanghai ： Shanghai Ocean University（吴志高. 基于改进YOLOv7的微藻轻量级检测算法研究与应用.上海：上海海洋大学）， 2023 .

Affonso C , Rossi A L D , Vieira F H A , Carvalho A C P D L F D . Expert Syst Appl , 2017 , 85 : 114 - 122 .

Alzubaidi L , Zhang J , Humaidi A J , AI-Dujaili A , Duan Y , AI-Shamma O , Santamaria J . J. Big Data , 2021 , 8 : 1 - 74 .

Redmon J , Divvala S , Girshick R , Farhadi A . Proc . IEEE Conf. Comput. Vis. Pattern Recog. . 2016 : 779 - 788 .

Park J , Baek J , You K , Nam S W , Kim J . Journal of KSWE , 2021 , 37 ( 4 ): 275 - 285 .

Park J , Baek J , Kim J , You K , Kim K . Water , 2022 , 14 ( 8 ): 1275 .

Su A . Research on portable microscopic system for field detection of microalgae . Jinan ： Shandong University （苏昂.用于微藻现场检测的便携式显微系统研究.济南：山东大学）， 2022 .

Wang L C ， Chen C J ， Zou Y W ， Xie T . J. Electron. Meas. Instrum. （汪鲁才，陈春江，邹伊雯，谢婷.电子测量与仪器学报）， 2024 , 38 ( 04 )： 46 - 54 .

Sunkara R , Luo T . ECML-PKDD , 2022 : 443 - 459 .

Li H , Li J , Wei H , Liu Z , Zhan Z , Ren Q . arxiv preprint arxiv : 2206 . 02424 , 2022 , 10 .

Li C Y , Li L L , Jiang H L , Weng K H , Geng Y F , Li L , Ke Z D , Li Q Y , Cheng M , Nie W Q , Li Y D , Zhang B , Liang Y F , Zhou L Y , Xu X M , Chu X X , Wei X M , Wei X L . arxiv preprint arxiv : 2209.02976 , 2022 .

He K , Zhang X , Ren S , Sun J . IEEE Trans. Pattern Anal. Mach. Intell , 2015 , 37 ( 9 ): 1904 - 1916 .

Kasper-Eulaers M , Hahn N , Berger S , Sebulonsen T , Kummervold P E . Algorithms , 2021 , 14 ( 4 ): 114 .

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

基于深度残差网络和近红外光谱的煤矸石智能识别

高分辨质谱与深度学习视阈下的藏红花品质分析

基于深度学习与可见－近红外光谱的患腥黑穗病小麦籽粒分类研究