基于级联高分辨率网络的活性污泥显微图像分割方法

doi:10.3969/j.issn.2095-2198.2023.02.008

摘要
参考文献
相关文章
推荐阅读
Metrics

下载:

PDF (4098KB)
输出: BibTeX | EndNote (RIS)

摘要

由于活性污泥显微图像固有的光晕、阴影等伪影以及絮状物和丝状菌多样性和结构性的不均匀性，传统图像分割方法存在絮体和丝状菌欠分割和过分割的问题.笔者基于高分辨率网络（high resolution network,HRNet）提出一种级联高分辨率网络（cascaded high resolution network,CHRNet）的活性污泥显微图像分割方法.该方法基于HRNet网络框架，通过特征金字塔构建级联的解码器结构，利用低分辨率提取语义信息，并中高分辨率恢复与细化边缘信息.引入多标签监督使反向传播更加平滑，从而实现更准确的预测.通过引入加权交叉熵损失函数，改善活性污泥显微图像分割中絮体和丝状菌样本类别不均衡的问题.真实污水处理厂活性污泥显微图像数据集图像分割实验结果表明：CHRNet方法在鲁棒性和泛化能力方面好于HRNet、U-Net和DeepLabV3+方法.

	服务
	把本文推荐给朋友
	加入引用管理器
	E-mail Alert
	RSS
	（）
	作者相关文章

关键词: 污水处理')" href="#">

污水处理活性污泥图像分割高分辨率表征级联结构

Abstract:

Due to the inherent microscope artifacts such as halos,shadows,and the diverse and uneven structures of flocs and filamentous bacteria in activated sludge microscopic images,traditional image segmentation methods have problems of under segmentation and over segmentation of flocs and filamentous bacteria.The author proposes a cascaded high resolution network (CHRNet) based activated sludge microscopic image segmentation method based on high resolution network (HRNet)The method utilizes the HRNet framework and constructs a cascaded decoder structure with a feature pyramid.By using low-resolution data for semantic information extraction and restoring/refining edge information using medium and high resolution data,CHRNet achieves more accurate predictions.In order to address the problem of class imbalance between flocs and filamentous bacteria,we introduce a weighted cross-entropy loss function.Experimental results on an activated sludge microscopic image dataset from a real waste water treatment plant show that CHRNet outperforms HRNet,U-Net,and DeepLabV3+ methods in terms of robustness and generalization ability.

Key words: waste water treatment')" href="#">

waste water treatment activated sludge image segmentation high-resolution representation cascade structure

出版日期: 2023-04-29 发布日期: 2024-06-06 整期出版日期: 2023-04-29

ZTFLH:

TP391

基金资助:

国家重点研发计划项目（2018YFB1700200）；2020年辽宁省高等学校创新人才支持计划、2021年高等学校基本科研项目重点项目（LJKZ0442）资助

引用本文:

赵立杰, 路星奎, 陈斌.

基于级联高分辨率网络的活性污泥显微图像分割方法 [J]. 沈阳化工大学学报, 2023, 37(2): 144-150.
ZHAO Lijie, LU Xingkui, CHEN Bin.

Microscopic Image Segmentation Method of Activated Sludge Based on Cascaded High-Resolution Network . Journal of Shenyang University of Chemical Technology, 2023, 37(2): 144-150.

链接本文:

https://xuebao.syuct.edu.cn/CN/10.3969/j.issn.2095-2198.2023.02.008 或 https://xuebao.syuct.edu.cn/CN/Y2023/V37/I2/144

［1］张秋卓,徐亚同,周扬,等.废水处理与水环境保护［J］.净水技术,2010，29(4):1-4.

［2］MOLINA M A,PREZ C A A,LEIVA C A.Characterization of Filamentous Flocs to Predict Sedimentation Parameters Using Image Analysis ［J］.Journal of Sensors,2020,2020:5248509.

［3］KHAN M B,NISAR H,NG C A,et al.Segmentation Approach Towards Phase-Contrast Microscopic Images of Activated Sludge to Monitor the Wastewater Treatment ［J］.Microscopy and Microanalysis,2017,23(6):1130-1142.

［4］CENENS C,VAN BEURDEN K P,JENN R,et al.On the Development of A Novel Image Analysis Technique to Distinguish between Flocs and Filaments in Activated Sludge Images ［J］.Water Science& Technology,2002,46(1/2):381-387.

［5］SHELHAMER E,LONG J,DARRELL T.Fully Convolutional Networks for Semantic Segmentation ［J］.IEEE Transactions on Pattern Analysis and Machine Intelligence,2017,39(4):640-651.

［6］ZHAO L J,ZOU S D,ZHANG Y H,et al.Segmentation of Activated Sludge Phase Contrast Microscopy Images Using U-Net Deep Learning Model ［J］.Sensors and Materials,2019,31(6):2013-2028.

［7］CORDTS M,OMRAN M,RAMOS S,et al.The Cityscapes Dataset for Semantic Urban Scene Understanding［C］//2016 IEEE Conference on Computer Vision and Pattern Recognition(CVPR).Las Vegas:IEEE,2016:3213-3223.

［8］NEWELL A,YANG K Y,DENG J.Stacked Hourglass Networks for Human Pose Estimation［C］//Computer Vision——ECCV 2016.Cham：Springer,2016:483-499.

［9］RONNEBERGER O,FISCHER P,BROX T.U-Net:Convolutional Networks for Biomedical Image Segmentation ［C］//Medical Image Computing and Computer-Assisted Intervention——MICCAI 2015.Cham:Springer,2015:234-241.

［10］ZHOU Z W,RAHMAN SIDDIQUEE M M,TAJBAKHSH N,et al.UNet++:A Nested U-Net Architecture for Medical Image Segmentation ［C］//Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support.Cham:Springer,2018:3-11.

［11］CHEN L C,PAPANDREOU G,KOKKINOS I,et al.DeepLab:Semantic Image Segmentation with Deep Convolutional Nets,Atrous Convolution,and Fully Connected CRFs ［J］.IEEE Transactions on Pattern Analysis and Machine Intelligence,2018,40(4):834-848.

［12］SUN K,XIAO B,LIU D,et al.Deep High-Resolution Representation Learning for Human Pose Estimation［C］//2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition（CVPR）Long Beach:IEEE,2019:5686-5696.

［13］LIN T Y,DOLLR P,GIRSHICK R,et al.Feature Pyramid Networks for Object Detection［C］//2017 IEEE Conference on Computer Vision and Pattern Recognition（CVPR）Honolulu:IEEE,2017:936-944.

［14］ZHAO H S,QI X J,SHEN X Y,et al.ICNet for Real-Time Semantic Segmentation on High-Resolution Images［C］//Computer Vision——ECCV 2018Cham：Springer,2018:418-434.

［15］LIU L,OUYANG W L,WANG X G,et al.Deep Learning for Generic Object Detection:A Survey ［J］.International Journal of Computer Vision,2020,128(2):261-318.

［16］UHL J H,LEYK S,CHIANG Y Y,et al.Spatialising Uncertainty in Image Segmentation Using Weakly Supervised Convolutional Neural Networks:A Case Study from Historical Map Processing ［J］.IET Image Processing,2018,12(11):2084-2091.

[1]	赵立杰, 吴志豪, 黄明忠. 基于SIFT特征匹配的活性污泥显微图像拼接方法 [J]. 沈阳化工大学学报, 2023, 37(1): 74-79.

No Suggested Reading articles found!

Viewed

Full text

Abstract

Cited

Shared

Discussed