基于集成学习的虹膜分割算法

doi:10.16180/j.cnki.issn1007-7820.2025.03.012

摘要/Abstract

摘要：

针对虹膜图像在分割过程中细节分割不准确、边界分割不圆滑以及易受噪声影响等问题,文中提出了一种基于集成学习的虹膜分割算法。相较于传统集成学习算法,文中算法基于皮尔森系数法选择合适的模型作为基学习器,从而提高了集成学习的性能。选用U²-Net、DeepLabv3+以及PSPNet作为同质个体学习器在CASIA-Iris-Interval数据集上进行训练,并预测得到对应的虹膜分割预测结果。对预测结果进行CLAHE和Gamma校正等图像增强操作得到新的预测结果图,选取加权平均法作为集成算法将基学习器的预测结果进行集成学习,从而得到最终的虹膜分割预测结果。测试结果表明,在3个不同的评估指标下,相较于基学习器,所提算法的准确率提升了1%,平均交并比提升了3.8%,宏平均分数提升了2.4%,视觉效果和客观评价指标均有所提升。

关键词: 虹膜分割, 集成学习, 学习器, U²-Net, DeepLabv3+, PSPNet, 皮尔森系数法, 图像处理

Abstract:

In view of the problems such as inaccurate detail segmentation, unsmooth boundary segmentation and easy to be affected by noise, an iris segmentation algorithm based on ensemble learning is proposed in this study. Compared with the traditional ensemble learning algorithm, the appropriate model is selected as the base learner based on Pearson coefficient method, so as to improve the performance of ensemble learning. U²-Net, DeepLabv3+ and PSPNet are selected as homogeneous individual learners to train on the CASIA-Iris-Interval dataset, and the corresponding iris segmentation prediction results are obtained. CLAHE and Gamma correction and other image enhancement operations are performed to obtain a new prediction result graph. Weighted average method is selected as an integrated algorithm to integrate the prediction results of the basis learner, so as to obtain the final prediction results of iris segmentation. The test results show that the accuracy of the proposed algorithm is improved by 1%, the average crossover ratio is improved by 3.8%, the average macro score is improved by 2.4%, and the visual effect and objective evaluation index have better segmentation effect when compared with the base learner under three different evaluation indexes.

Key words: iris egmentation, integrated learning, learner, U²-Net, DeepLabv3+, PSPNet, Pearson coefficient method, image processing

中图分类号:

TP391.4

孙佳倩, 朱金荣, 张小宝, 张云恺, 龚卫娟. 基于集成学习的虹膜分割算法[J]. 电子科技, 2025, 38(3): 88-94.

SUN Jiaqian, ZHU Jinrong, ZHANG Xiaobao, ZHANG Yunkai, GONG Weijuan. Research on Iris Segmentation Algorithm Based on Ensemble Learning[J]. Electronic Science and Technology, 2025, 38(3): 88-94.

图/表 11

表1

图1

图2

图3

图4

图5

图6

图7

图8

表2

表3

参考文献 18

[1]	Long J, Shelhamer E, Darreii T. Fully convolutional networks for semantic segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015, 39(4):640-651.
[2]	Lian S, Luo Z, Zhong Z, et al. Attention guided U-Net for accurate iris segmentation[J]. Journal of Visual Communication and Image Representation, 2018, 56(1):296-304.
[3]	Orlando I J, Seeböck P, Bogunovic H, et al. U²-Net:A Bayesian U-Net model with epistemic uncertainty feedback for photoreceptor layer segmentation in pathological OCT scans[J]. Computing Research Repository, 2019, 10(9):123-128.
[4]	欧晓焱. 基于注意力机制的DeepLab v3+语义分割算法研究[D]. 南京: 南京邮电大学,2022:1-42.
	Ou Xiaoyan. Research on DeepLab v3+ semantic segmentation algorithm based on attention mechanism[D]. Nanjing: Nanjing University of Posts and Telecommunications, 2022:1-42.
[5]	王盛, 吴浩, 彭宁, 等. 改进U²-Net的太阳能电池片缺陷分割方法[J]. 国外电子测量技术, 2023, 42(2):177-184.
	Wang Sheng, Wu Hao, Peng Ning, et al. Improved U²-Net defect segmentation method for solar cells[J]. Foreign Electronic Measurement Technology, 2023, 42(2):177-184.
[6]	陈其浩, 孙林, 张倩. 基于改进U²-Net的透明件划痕检测方法[J]. 科学技术与工程, 2022, 22(2):620-627.
	Chen Qihao, Sun Lin, Zhang Qian. Scratch detection method of transparent parts based on improved U²-Net[J]. Science Technology and Engineering, 2022, 22(2):620-627.
[7]	胡智翔, 鲍胜利, 徐传淇, 等. 基于DeepLab v3+的遥感建筑物语义分割算法[J]. 计算机应用, 2021, 41(S2):71-75.
	Hu Zhixiang, Bao Shengli, Xu Chuanqi, et al. Semantic segmentation algorithm for remote sensing buildings based on DeepLab v3+[J]. Journal of Computer Applications, 2021, 41(S2):71-75.
[8]	杨中科. 基于深度卷积神经网络的图像语义分割算法研究[D]. 太原: 太原理工大学,2021:1-52.
	Yang Zhongke. Research on image semantic segmentation algorithm based on deep convolutional neural networks[D]. Taiyuan: Taiyuan University of Technology,2021:1-52.
[9]	徐聪, 王丽. 基于改进DeepLabv3+网络的图像语义分割方法[J]. 激光与光电子学进展, 2021, 58(16):225-232.
	Xu Cong, Wang Li. Imagesemantic segmentation method based on improved DeepLab v3+ network[J]. Laser and Optoelectronics Progress, 2021, 58(16):225-232.
[10]	张乔木, 钟倩文, 孙明, 等. 复杂环境下弓网接触位置动态监测方法研究[J]. 电子科技, 2022, 35(8):66-72.
	Zhang Qiaomu, Zhong Qianwen, Sun Ming, et al. Research on dynamic monitoring method of pantograph-net contact position in complex environment[J]. Electronic Science and Technology, 2022, 35(8):66-72.
[11]	罗会兰, 易慧. 基于迭代训练和集成学习的图像分类方法[J]. 计算机工程与设计, 2020, 41(5):1301-1307.
	Luo Huilan, Yi Hui. Image classification method based on iterative training and ensemble learning[J]. Computer Engineering and Design, 2020, 41(5):1301-1307.
[12]	闫超, 孙占全, 田恩刚, 等. 基于深度学习的医学图像分割技术研究进展[J]. 电子科技, 2021, 34(2):7-11.
	Yan Chao, Sun Zhanquan, Tian Engang, et al. Research progress of medical image segmentation based on deep learning[J]. Electronic Science and Technology, 2021, 34(2):7-11.
[13]	蒋芸, 王发林, 张海. 基于集成分类型深度神经网络的视网膜眼底血管图像分割[J]. 计算机工程与科学, 2021, 43(5):862-871.
	Jiang Yun, Wang Falin, Zhang Hai. Image segmentation of retinal fundus vessels based on ensembled classified deep neural network[J]. Computer Engineering and Science, 2021, 43(5):862-871.
[14]	朱诗生, 王慧娟, 李淳鑫. 基于深度学习和模型集成的肺结节分割方法[J]. 计算机技术与发展, 2023, 33(2):208-213.
	Zhu Shisheng, Wang Huijuan, Li Chunxin. Pulmonary nodule segmentation method based on deep learning and model integration[J]. Computer Technology and Development, 2023, 33(2):208-213.
[15]	王财勇, 孙哲南. 虹膜分割算法评价基准[J]. 计算机研究与发展, 2020, 57(2):395-412.
	Wang Caiyong, Sun Zhenan. A benchmark for iris segmentation[J]. Journal of Computer Research and Development, 2020, 57(2):395-412.
[16]	尤轩昂, 赵鹏, 慕晓冬, 等. 融合注意力机制与密集多尺度特征的异质噪声虹膜分割方法[J]. 激光与光电子学进展, 2022, 59(4):109-120.
	You Xuanang, Zhao Peng, Mu Xiaodong, et al. Heterogeneousnoise iris segmentation based on attention mechanism and dense multiscale features[J]. Laser and Optoelectronics Progress, 2022, 59(4):109-120.
[17]	张珅睿. 基于神经网络的图像语义分割方法[D]. 武汉: 华中科技大学,2021:1-68.
	Zhang Shenrui. Image semantic segmentation method based on neural network[D]. Wuhan: Huazhong University of Science and Technology,2021:1-68.
[18]	李艾瑾. 基于注意力机制和深度多尺度融合网络的语义分割研究[D]. 西安: 西安电子科技大学,2021:1-64.
	Li Aijin. Attention mechanism and deep multiscale fusion network for semantic segmentation[D]. Xi'an: Xidian University,2021:1-64.

模型	U²-Net	DeepLabv3+	PSPNet
U²-Net	1.00	0.69	0.72
DeepLabv3+	0.69	1.00	0.85
PSPNet	0.72	0.85	1.00

方法	Acc/%	MIoU/%	mF1/%
U²-Net	96.8	93.4	96.5
DeepLabv3+	97.4	93.2	96.4
PSPNet	97.8	94.5	97.1
集成学习	97.8	96.7	98.8

方法	Acc/%	MIoU/%	mF1/%
U²-Net	97.8	93.9	97.4
DeepLabv3+	96.8	93.1	96.3
PSPNet	97.6	94.7	97.3
集成学习	97.8	96.9	97.7