基于双分支残差网络的粒子图像增强方法

张志浩, 于长东, 刘百胜, 范毅伟

doi:10.16411/j.cnki.issn1006-7736.2024.04.011

大连海事大学学报 >

2024 , Vol. 50 >Issue 4: 100 - 109

DOI: https://doi.org/10.16411/j.cnki.issn1006-7736.2024.04.011

基于双分支残差网络的粒子图像增强方法

展开

(1.大连海事大学人工智能学院, 辽宁大连, 116026; 2.辽宁工程技术大学软件学院，辽宁葫芦岛 125000;3.哈尔滨工程大学船舶工程学院，哈尔滨 150001)

张志浩(2000 - )，男，研究生；于长东*(1996 - )，男，博士，讲师，E-mail：ycd@dlmu.edu.cn

网络出版日期: 2024-05-24

基金资助

国家自然科学基金青年基金资助项目(52401362)；中央高校基本科研业务费专项资金资助项目(3132024224)

收起

Particle image enhancement method based on dual-branch residual network

Expand

(1. College of Artificial Intelligence, Dalian Maritime University, Dalian 116026, China; 2. College of Software, Liaoning Technical University，Huludao 125000, China;3. College of Shipbuilding Engineering, Harbin Engineering University, Harbin 150001, China )

Online published: 2024-05-24

Fold

摘要

为获取高质量的粒子图像，提出一种双分支残差卷积神经网络DBRNet用于粒子图像测速(PIV)技术中的粒子图像增强。首先，设计一种由残差块组成的双分支卷积神经网络对输入的粒子图像对进行特征提取，同时，采用编码-解码器对粒子图像对的特征信息进行有效融合。其次，自主生成具有挑战性的图像增强数据集训练模型参数，其中，包含不同浓度的高斯噪声、光强噪声及多种真实的干扰背景，以充分模拟真实流体场景。结果表明，本文方法能够有效地处理合成图像和真实图像中的噪声干扰，实现图像增强；利用速度场估计算法处理经本文方法增强后的粒子图像对可以得到更高精度的速度场。

关键词： 粒子图像增强; 粒子图像测速(PIV); 深度学习; 双分支残差网络

本文引用格式

张志浩, 于长东, 刘百胜, 范毅伟 . 基于双分支残差网络的粒子图像增强方法[J]. 大连海事大学学报, 2024 , 50(4) : 100 -109 . DOI: 10.16411/j.cnki.issn1006-7736.2024.04.011

Abstract

A dual-branch residual convolutional neural network was proposed for image enhancement in PIV velocity technique to obtain high-quality particle images. Firstly, a dual-branch convolutional neural network composed of residual blocks was designed to extract features from the input particle image pairs, while a coding-decoder was used to effectively fuse the feature information of the particle image pairs. Secondly, a challenging image enhancement dataset was autonomously generated to train model parameters, including Gaussian noises of different concentrations, light intensity noise and various real interference backgrounds, thereby fully simulating real fluid scenes. Results show that the proposed method can effectively deal with noise interference in both synthesized and real images, achieving image enhancement. Meanwhile， higher precision velocity fields can be obtained by using velocity field estimation algorithm to process the particle image pairs enhanced by the proposed method in this paper.

Key words： particle image enhancement; particle image velocimetry(PIV); deep learning; dual-branch residual network

参考文献

[1] WILLERT C E, GHARIB M. Digital particle image velocimetry[J]. Experiments in Fluids, 1991, 10: 181-193.
[2] YU C D, CHANG Y P, LIANG X, et al. Deep learning for particle image velocimetry with attentional transformer and cross-correlation embedded[J]. Ocean Engineering, 2024, 292: 116522.
[3] OZAWA Y, HONDA H, NONOMURA T. Spatial super resolution based on simultaneous dual PIV measurement with different magnification[J]. Experiments in Fluids, 2024, 65: 42.
[4] 毕晓君,何明洁,于长东, 等.基于深度学习的液相流粒子图像测速估计[J].哈尔滨工程大学学报, 2023, 44(4):622-630.
BI X J, HE M J, YU C D, et al. Particle image velocimetry for liquid phase flow based on deep learning [J]. Journal of Harbin Engineering University, 2023, 44(4):622-630. (in Chinese)
[5] WENG W G, FAN W C, LIAO G X, et al. Wavelet-based image denoising in (digital) particle image velocimetry[J]. Signal Processing, 2001, 81(7): 1503-1512.
[6]DE LIMA AMARAL R, BORTOLIN V A A, LEMOS B L H D, et al. A novel method based on the Otsu threshold for instantaneous elimination of light reflection in PIV images[J]. Measurement Science and Technology, 2021, 33(2): 025401.
[7] STANISLAS M, OKAMOTO K, KÄHLER C. Main results of the first international PIV challenge[J]. Measurement Science and Technology, 2003, 14(10): R63.
[8] SCIACCHITANO A, SCARANO F. Elimination of PIV light reflections via a temporal high pass filter[J]. Measurement Science and Technology, 2014, 25(8): 084009.
[9] HONKANEN M, NOBACH H. Background extraction from double-frame PIV images[J]. Experiments in Fluids, 2005, 38: 348-362.
[10] DEEN N G, WILLEMS P, VAN SINT ANNALAND M, et al. On image pre-processing for PIV of single-and two-phase flows over reflecting objects[J]. Experiments in Fluids, 2010, 49: 525-530.
[11] MENDEZ M A, RAIOLA M, MASULLO A, et al. POD-based background removal for particle image velocimetry[J]. Experimental Thermal and Fluid Science, 2017, 80: 181-192.
[12] ADATRAO S, SCIACCHITANO A. Elimination of unsteady background reflections in PIV images by anisotropic diffusion[J]. Measurement Science and Technology, 2019, 30(3): 035204.
[13] CAI S Z, ZHOU S C, XU C, et al. Dense motion estimation of particle images via a convolutional neural network[J]. Experiments in Fluids, 2019, 60: 1-16.
[14] 于长东, 毕晓君, 韩阳, 等.基于轻量化深度学习模型的粒子图像测速研究[J]. 光学学报, 2020, 40(7):142-149.
YU C D, BI X J, HAN Y, et al. Particle image velocimetry based on a lightweight deep learning model [J]. Acta Optica Sinica, 2020, 40(7):142-149. (in Chinese)
[15] YU C D, BI X J, FAN Y W, et al. LightPIVNet: An effective convolutional neural network for particle image velocimetry[J]. IEEE Transactions on Instrumentation and Measurement, 2021, 70: 1-15.
[16] FAN Y W, GUO C Y, HAN Y, et al. Deep-learning-based image preprocessing for particle image velocimetry[J]. Applied Ocean Research, 2023, 130: 103406.
[17] HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. New York: IEEE,2016: 770-778.
[18] IOFFE S, SZEGEDY C. Batch normalization: accelerating deep network training by reducing internal covariate shift[C]//International Conference on Machine Learning. PMLR, 2015: 448-456.
[19] ZHANG K, ZUO W M, CHEN Y J, et al. Beyond a gaussian denoiser: Residual learning of deep CNN for image denoising[J]. IEEE Transactions on Image Processing, 2017, 26(7): 3142-3155.
[20] ADRIAN R J. Particle-imaging techniques for experimental fluid mechanics[J]. Annual Review of fluid Mechanics, 1991, 23(1): 261-304.
[21] LEE Y, YANG H, YIN Z P. PIV-DCNN: cascaded deep convolutional neural networks for particle image velocimetry[J]. Experiments in Fluids, 2017, 58: 1-10.
[22] THIELICKE W, STAMHUIS E J. PIVlab–towards user-friendly, affordable and accurate digital particle image velocimetry in MATLAB[J]. Journal of Open Research Software, 2014, 2: 30.
[23]WANG C X, JIN Z. Brighten-and-colorize: A decoupled network for customized low-light image enhancement[C]//Proceedings of the 31st ACM International Conference on Multimedia. 2023: 8356-8366.
[24] SCARANO F. Iterative image deformation methods in PIV[J]. Measurement Science and Technology, 2001, 13(1): R1.
[25] WESTERWEEL J, SCARANO F. Universal outlier detection for PIV data[J]. Experiments in Fluids, 2005, 39: 1096-1100.

Options

文章导航

模态框（Modal）标题

摘要

本文引用格式

Abstract

参考文献