基于THGS算法优化ResNet-18模型的图像识别

doi:10.13229/j.cnki.jdxbgxb.20230775

摘要/Abstract

摘要：

为快速、准确识别水稻褐斑病图像，提出一种改进的THGS-ResNet-18识别模型。首先，应用Tent混沌映射改进饥饿游戏搜索（Hunger game search， HGS）算法，解决HGS算法种群初始化随机性过大的问题；其次，应用改进后的HGS算法优化ResNet-18模型的超参数；最后，应用改进模型THGS-ResNet-18针对5064张水稻叶片图像进行识别，且与经过其他4个群体智能算法优化的ResNet-18模型的7个评价指标进行了比较。实验表明，相较于其他4种算法，本文所提算法优化模型的准确率提升了5.22~6.09百分点，敏感性提升了3.53~5.31百分点，特异性提升了7.38百分点，精度提升了6.95~7.13百分点，召回率提升了3.53~5.31百分点，F-measure提升了5.22~6.20百分点，G-mean提升了5.24~6.13百分点。

关键词: 深度学习, 饥饿游戏搜索算法, 混沌映射, 超参数优化, 水稻病害

Abstract:

This article proposes an improved THGS ResNet-18 recognition model for fast and accurate recognition of rice brown spot images. Firstly， apply Tent chaotic mapping to improve the hunger game search （HGS） algorithm， solving the problem of excessive randomness in the population initialization of the HGS algorithm. Secondly， the improved HGS algorithm hyperparameter is applied to optimize ResNet-18 model. Finally， the improved model THGS ResNet-18 was used to recognize 5 064 rice leaf images， and compared with other four ResNet-18 models improved by swarm intelligence algorithm for seven evaluation indicators. Experiments showed that the accuracy rate of the model proposed in this paper increased by 5.22~6.09 percentage points， sensitivity by 3.53~5.31 percentage points， specificity by 7.38 percentage points， precision by 6.95~7.13 percentage points， recall rate by 3.53~5.31 percentage points， f-measure by 5.22~6.20 percentage points， and g-mean by 5.24~6.13 percentage points.

Key words: deep learning, hunger game search algorithm, chaotic mapping, hyperparameter optimization, rice diseases

中图分类号:

TP399

李健,刘欢,李艳秋,王海瑞,关路,廖昌义. 基于THGS算法优化ResNet-18模型的图像识别[J]. 吉林大学学报(工学版), 2025, 55(5): 1629-1637.

Jian LI,Huan LIU,Yan-qiu LI,Hai-rui WANG,Lu GUAN,Chang-yi LIAO. Image recognition research on optimizing ResNet-18 model based on THGS algorithm[J]. Journal of Jilin University(Engineering and Technology Edition), 2025, 55(5): 1629-1637.

图/表 10

图1

图2

图3

图4

图5

图6

表1

表2

图7

图8

参考文献 22

[1]	Chen Y S, Wang Y, Gu Y F, et al. Deep learning ensemble for hyperspectral image classification [J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2019, 12(6): 1882-1897.
[2]	Lee D, Lee J, Ko J, et al. Deep learning in MR image processing [J]. Investigative Magnetic Resonance Imaging, 2019, 23(2): 81-99.
[3]	Kwon D, Kim H, Kim J, et al. A survey of deep learning-based network anomaly detection [J]. Cluster Computing-the Journal of Networks Software Tools and Applications, 2019, 22: 949-961.
[4]	Hsieh T H, Kiang J F. Comparison of CNN algorithms on hyperspectral image classification in agricultural lands [J]. Sensors, 2020, 20(6): s20061734.
[5]	Hua S, Xu M J, Xu Z F, et al. Multi-feature decision fusion algorithm for disease detection on crop surface based on machine vision [J]. Neural Computing & Applications, 2022, 34(12): 9471-9484.
[6]	Sathya K, Rajalakshmi M. RDA- CNN: enhanced super resolution method for rice plant disease classification [J]. Computer Systems Science and Engineering, 2022, 42(1): 33-47.
[7]	Wang Y B, Wang H F, Peng Z H. Rice diseases detection and classification using attention based neural network and bayesian optimization [J]. Expert Systems with Applications, 2021, 178: 114770.
[8]	Chen J Y, Lin X, Gao S T D, et al. A fast evolutionary learning to optimize CNNinspec keywordsother keywordskey words [J]. Chinese Journal of Electronics, 2020, 29(6): 1061-1073.
[9]	Chen K C, Huang Y W, Liu G M, et al. A hierarchical k-means-assisted scenario-aware reconfigurable convolutional neural network [J]. IEEE Transactions on Very Large Scale Integration Systems, 2021, 29(1): 176-188.
[10]	Song Y, He B, Liu P. Real-time object detection for auvs using self-cascaded convolutional neural networks [J]. IEEE Journal of Oceanic Engineering, 2021, 46(1): 56-67.
[11]	刘培勇,董洁,谢罗峰,等. 基于多支路卷积神经网络的磁瓦表面缺陷检测算法[J].吉林大学学报: 工学版, 2023, 53(5): 1449-1457.
	Liu Pei-yong, Dong Jie, Xie Luo-feng, et al. Magnetic tile surface defect detection algorithm based on multi-branch convolutional neural network[J]. Journal of Jilin University (Engineering and Technology Edition), 2023, 53(5): 1449-1457.
[12]	Yang Y T, Chen H L, Heidari A A, et al. Hunger games search: visions, conception, implementation, deep analysis, perspectives, and towards performance shifts [J]. Expert Systems with Applications, 2021, 177: 114864.
[13]	Zhang Y D, Mo Y B. Chaotic adaptive sailfish optimizer with genetic characteristics for global optimization [J]. Journal of Supercomputing, 2022, 78(8): 10950-10996.
[14]	Ma J, Hao Z Y, Sun W J. Enhancing sparrow search algorithm via multi-strategies for continuous optimization problems [J]. Information Processing & Management, 2022, 59(2): 102854.
[15]	Tsuneda A. Orthogonal chaotic binary sequences based on tent map and walsh functions [J]. Ieice Transactions on Fundamentals of Electronics Communications and Computer Sciences, 2021, 104(9): 1349-1352.
[16]	Valle J, Machicao J, Bruno O M. Chaotical PRNG based on composition of logistic and tent maps using deep-zoom [J]. Chaos Solitons & Fractals, 2022, 161: 112296.
[17]	Liu H, Li J, Du J, et al. Identification of smoke from straw burning in remote sensing images with the improved yolov5s algorithm [J]. Atmosphere, 2022, 13(6): 13060925.
[18]	Huang Y, Yu K, Wu N, et al. Slope shape and edge intelligent recognition technology based on deep neural sensing network [J]. 2022, 01: 5901803.
[19]	Zhang Y Q, Peng L X, Ma G L, et al. Dynamic gesture recognition model based on millimeter-wave radar with ResNet-18 and LSTM [J]. Frontiers in Neurorobotics, 2022, 16: 909137.
[20]	杨怀江,王二帅,隋永新,等. 简化型残差结构和快速深度残差网络[J].吉林大学学报: 工学版, 2022, 52(6): 1413-1421.
	Yang Huai-jiang, Wang Er-shuai, Sui Yong-xin, et al. Simplified residual structure and fast deep residual network [J]. Journal of Jilin University (Engineering and Technology Edition), 2022, 52(6): 1413-1421.
[21]	Chen X. Vehicle feature recognition via a convolutional neural network with an improved bird swarm algorithm [J]. Journal of Internet Technology, 2023, 24(2): 421-432.
[22]	He X X, Shan W F, Zhang R L, et al. Improved colony predation algorithm optimized convolutional neural networks for electrocardiogram signal classification [J]. Biomimetics, 2023, 8(3): 8030268.

相关文章 15

[1]	文斌,丁弈夫,杨超,沈艳军,李辉. 基于自选择架构网络的交通标志分类算法[J]. 吉林大学学报(工学版), 2025, 55(5): 1705-1713.
[2]	张汝波,常世淇,张天一. 基于深度学习的图像信息隐藏方法综述[J]. 吉林大学学报(工学版), 2025, 55(5): 1497-1515.
[3]	李振江,万利,周世睿,陶楚青,魏巍. 基于时空Transformer网络的隧道交通运行风险动态辨识方法[J]. 吉林大学学报(工学版), 2025, 55(4): 1336-1345.
[4]	赵孟雪,车翔玖,徐欢,刘全乐. 基于先验知识优化的医学图像候选区域生成方法[J]. 吉林大学学报(工学版), 2025, 55(2): 722-730.
[5]	刘元宁,臧子楠,张浩,刘震. 基于深度学习的核糖核酸二级结构预测方法[J]. 吉林大学学报(工学版), 2025, 55(1): 297-306.
[6]	徐慧智,蒋时森,王秀青,陈爽. 基于深度学习的车载图像车辆目标检测和测距[J]. 吉林大学学报(工学版), 2025, 55(1): 185-197.
[7]	李路,宋均琦,朱明,谭鹤群,周玉凡,孙超奇,周铖钰. 基于RGHS图像增强和改进YOLOv5网络的黄颡鱼目标提取[J]. 吉林大学学报(工学版), 2024, 54(9): 2638-2645.
[8]	张磊,焦晶,李勃昕,周延杰. 融合机器学习和深度学习的大容量半结构化数据抽取算法[J]. 吉林大学学报(工学版), 2024, 54(9): 2631-2637.
[9]	乔百友,武彤,杨璐,蒋有文. 一种基于BiGRU和胶囊网络的文本情感分析方法[J]. 吉林大学学报(工学版), 2024, 54(7): 2026-2037.
[10]	郭昕刚,何颖晨,程超. 抗噪声的分步式图像超分辨率重构算法[J]. 吉林大学学报(工学版), 2024, 54(7): 2063-2071.
[11]	张丽平,刘斌毓,李松,郝忠孝. 基于稀疏多头自注意力的轨迹kNN查询方法[J]. 吉林大学学报(工学版), 2024, 54(6): 1756-1766.
[12]	孙铭会,薛浩,金玉波,曲卫东,秦贵和. 联合时空注意力的视频显著性预测[J]. 吉林大学学报(工学版), 2024, 54(6): 1767-1776.
[13]	陆玉凯,袁帅科,熊树生,朱绍鹏,张宁. 汽车漆面缺陷高精度检测系统[J]. 吉林大学学报(工学版), 2024, 54(5): 1205-1213.
[14]	李雄飞,宋紫萱,朱芮,张小利. 基于多尺度融合的遥感图像变化检测模型[J]. 吉林大学学报(工学版), 2024, 54(2): 516-523.
[15]	杨国俊,齐亚辉,石秀名. 基于数字图像技术的桥梁裂缝检测综述[J]. 吉林大学学报(工学版), 2024, 54(2): 313-332.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed

指标	数值	指标	数值
ACC	96.05	Recall	94.60
Sensitivity	94.60	F-measure	95.89
Specificity	97.44	G-mean	96.01
Precision	97.22	-	-

模型指标	GA-ResNet-18	PSO-ResNet-18	GWO-ResNet-18	WOA-ResNet-18
ACC	89.96	90.39	90.39	90.83
Sensitivity	89.29	90.18	90.18	91.07
Specificity	90.60	90.60	90.60	90.60
Precision	90.09	90.18	90.18	90.27
Recall	89.29	90.18	90.18	91.07
F-measure	89.69	90.18	90.18	90.67
G-mean	89.94	90.39	90.39	90.83