基于卷积神经网络的生物式水质监测方法
|
摘要
生物式水质监测通常是先通过提取水生物在不同环境下的应激反应特征,再进行特征分类,从而识别水质。针对水质监测问题,提出一种使用卷积神经网络(CNN)的方法。鱼类运动轨迹是当前所有文献使用的多种水质分类特征的综合性表现,是生物式水质分类的重要依据。使用Mask-RCNN的图像分割方法,求取鱼体的质心坐标,并绘制出一定时间段内鱼体的运动轨迹图像,制作正常与异常水质下两种轨迹图像数据集。融合Inception-v3网络作为数据集的特征预处理部分,重新建立卷积神经网络对Inception-v3网络提取的特征进行分类。通过设置多组平行实验,在不同的水质环境中对正常水质与异常水质进行分类。结果表明,卷积神经网络模型的水质识别率为99. 38%,完全达到水质识别的要求。
Biological water quality monitoring is usually through the extraction of water stress response characteristics in different environments,and then feature classification,so as to identify water quality. Aiming at the problem of water quality monitoring,method of CNN convolution neural network was presented. Fish trajectory is a comprehensive expression of the various water quality classification characteristics used in all the literatures and is an important basis for the classification of biological water quality. Using the image segmentation method of Mask-RCNN to obtain the centroid coordinates of the fish and draw the trajectory image of the fish in a certain period of time. Two sets of trajectory image data sets under normal and abnormal water quality were produced. The Inception-v3 network serves as a feature preprocessing part of the data set,the CNN convolution neural network was reestablished to classify the features extracted by Inception-v3 network. Set up multiple sets of parallel experiments to classify normal and abnormal water quality in different environments. The results showed that the CNN convolution neural network model had a water quality recognition rate of99. 38%,which met the requirements of water quality identification.
Biological water quality monitoring is usually through the extraction of water stress response characteristics in different environments,and then feature classification,so as to identify water quality. Aiming at the problem of water quality monitoring,method of CNN convolution neural network was presented. Fish trajectory is a comprehensive expression of the various water quality classification characteristics used in all the literatures and is an important basis for the classification of biological water quality. Using the image segmentation method of Mask-RCNN to obtain the centroid coordinates of the fish and draw the trajectory image of the fish in a certain period of time. Two sets of trajectory image data sets under normal and abnormal water quality were produced. The Inception-v3 network serves as a feature preprocessing part of the data set,the CNN convolution neural network was reestablished to classify the features extracted by Inception-v3 network. Set up multiple sets of parallel experiments to classify normal and abnormal water quality in different environments. The results showed that the CNN convolution neural network model had a water quality recognition rate of99. 38%,which met the requirements of water quality identification.
引文
[1] Kim C M,Shin M W,Jeong S M,et al. Real-time motion generating method for artifical fish[J]. Computer Science and Network Security,2007,7(10):52-61.
[2] Lai C L,Chiu C L. Using image processing technology for water quality monitoring system[C]//ICMLC.International conference on machine learning and cybernetics. Guilin,China,2011:1856-1861.
[3] Zheng H Y,Liu R,Zhang R,et al. A method for realtime measurement of respiratory rhythms in medaka(Oryzias latipes)using computer vision for water quality monitoring[J]. Ecotoxicology and environmental safety,2014,100:76-86.
[4] Ma H,Tsai T F,Liu C C. Real-time monitoring of water quality using temporal trajectory of live fish[J]. Expert Systems with Applications,2010,37(7):5158-5171.
[5]程淑红,李雷华,刘洁,等.基于视觉感知的鱼群运动行为特征参数提取[J].计量学报,2017,38(2):175-178.Cheng S H, Li L H, Liu J, et al. Fish Movement Behavior Characteristic Parameter Extraction Based on Visual Perception[J]. Acta Metrologica Sinica,2017,38(2):175-178.
[6]程淑红,刘洁,李雷华.基于鱼类运动行为的水质异常评价因子研究[J].仪器仪表学报,2015,36(8):1759-1766.Cheng S H,Liu J,Li L H. Study on anomaly water quality assessment factor based on fish movement behavior[J]. Chinese Journal of scientific Instrument,2015,36(8):1759-1766.
[7] Cauwenberghs G. Incremental and decremental support vector machine[J]. Machine Learning,2001,44(13):409-415.
[8] Chen T Q, Guestrin C. XGBoost:A Scalable Tree Boosting System[C]//ACM. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York,USA,2016:785-794.
[9] Guo G D,Wang H,Bell D A,et al. KNN Model-Based Approach in Classification[J]. Lecture Notes in Computer Science,2003,2888:986-996.
[10]程淑红,刘洁.基于MWF和GF的复杂光照下人脸识别研究[J].计量学报,2017,38(1):60-64.Cheng S H,Liu J. Face Recognition under Complex Illumination Based on Multi-scale Weberface and Gradientface[J]. Acta Metrologica Sinica,2017,38(1):60-64.
[11]陈杰,尚丽.基于核竞争学习算法的图像特征提取[J].计量学报,2017,38(5):576-579.Chen J, Shang L. Image Feature Extraction Using Kernel Winner-take-all Based on Independent Component Analysis Algorithm[J]. Acta Metrologica Sinica,2017,38(5):576-579.
[12] Lin T Y,Dollár P,Girshick R,et al. Belongie,Feature Pyramid Networks for Object Detection[C]//IEEE. 2017 IEEE Conference on Computer Vision and Pattern Recognition(CVPR). Honolulu, HI,2017:936-944.
[13] Szegedy C,Liu W,Jia Y,et al. Going deeper with convolutions[C]//IEEE. 2015 IEEE Conference on Computer Vision and Pattern Recognition(CVPR).Boston,MA,2015:1-9.
[14] Ioffe S,Szegedy C. Batch Normalization:Accelerating Deep Network Training by Reducing Internal Covariate Shift[J]. Computer Science,2015,72(18):75-83.
[15] Szegedy C,Vanhoucke V,Ioffe S,et al. Rethinking the Inception Architecture for Computer Vision[C]//IEEE.2016 IEEE Conference on Computer Vision and Pattern Recognition(CVPR). Las Vegas, NV,2015:2818-2826.
[16] Xiao X F,Jin L W,Yang Y F,et al. Building fast and compact convolutional neural networks for offline handwritten Chinese character recognition[J]. Pattern Recognition,2017,72:72-81.
[17] Wilson D R,Martinez T R. The general inefficiency of batch training for gradient descent learning[J]. Neural Netw,2003,16(10):1429-1451.
[18] Cshalev-shwartz S, Singer Y. Primal Estimated Subgradient Solver for SVM[J]. Mathematical Programming,2011,127(1):3-30.
[2] Lai C L,Chiu C L. Using image processing technology for water quality monitoring system[C]//ICMLC.International conference on machine learning and cybernetics. Guilin,China,2011:1856-1861.
[3] Zheng H Y,Liu R,Zhang R,et al. A method for realtime measurement of respiratory rhythms in medaka(Oryzias latipes)using computer vision for water quality monitoring[J]. Ecotoxicology and environmental safety,2014,100:76-86.
[4] Ma H,Tsai T F,Liu C C. Real-time monitoring of water quality using temporal trajectory of live fish[J]. Expert Systems with Applications,2010,37(7):5158-5171.
[5]程淑红,李雷华,刘洁,等.基于视觉感知的鱼群运动行为特征参数提取[J].计量学报,2017,38(2):175-178.Cheng S H, Li L H, Liu J, et al. Fish Movement Behavior Characteristic Parameter Extraction Based on Visual Perception[J]. Acta Metrologica Sinica,2017,38(2):175-178.
[6]程淑红,刘洁,李雷华.基于鱼类运动行为的水质异常评价因子研究[J].仪器仪表学报,2015,36(8):1759-1766.Cheng S H,Liu J,Li L H. Study on anomaly water quality assessment factor based on fish movement behavior[J]. Chinese Journal of scientific Instrument,2015,36(8):1759-1766.
[7] Cauwenberghs G. Incremental and decremental support vector machine[J]. Machine Learning,2001,44(13):409-415.
[8] Chen T Q, Guestrin C. XGBoost:A Scalable Tree Boosting System[C]//ACM. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York,USA,2016:785-794.
[9] Guo G D,Wang H,Bell D A,et al. KNN Model-Based Approach in Classification[J]. Lecture Notes in Computer Science,2003,2888:986-996.
[10]程淑红,刘洁.基于MWF和GF的复杂光照下人脸识别研究[J].计量学报,2017,38(1):60-64.Cheng S H,Liu J. Face Recognition under Complex Illumination Based on Multi-scale Weberface and Gradientface[J]. Acta Metrologica Sinica,2017,38(1):60-64.
[11]陈杰,尚丽.基于核竞争学习算法的图像特征提取[J].计量学报,2017,38(5):576-579.Chen J, Shang L. Image Feature Extraction Using Kernel Winner-take-all Based on Independent Component Analysis Algorithm[J]. Acta Metrologica Sinica,2017,38(5):576-579.
[12] Lin T Y,Dollár P,Girshick R,et al. Belongie,Feature Pyramid Networks for Object Detection[C]//IEEE. 2017 IEEE Conference on Computer Vision and Pattern Recognition(CVPR). Honolulu, HI,2017:936-944.
[13] Szegedy C,Liu W,Jia Y,et al. Going deeper with convolutions[C]//IEEE. 2015 IEEE Conference on Computer Vision and Pattern Recognition(CVPR).Boston,MA,2015:1-9.
[14] Ioffe S,Szegedy C. Batch Normalization:Accelerating Deep Network Training by Reducing Internal Covariate Shift[J]. Computer Science,2015,72(18):75-83.
[15] Szegedy C,Vanhoucke V,Ioffe S,et al. Rethinking the Inception Architecture for Computer Vision[C]//IEEE.2016 IEEE Conference on Computer Vision and Pattern Recognition(CVPR). Las Vegas, NV,2015:2818-2826.
[16] Xiao X F,Jin L W,Yang Y F,et al. Building fast and compact convolutional neural networks for offline handwritten Chinese character recognition[J]. Pattern Recognition,2017,72:72-81.
[17] Wilson D R,Martinez T R. The general inefficiency of batch training for gradient descent learning[J]. Neural Netw,2003,16(10):1429-1451.
[18] Cshalev-shwartz S, Singer Y. Primal Estimated Subgradient Solver for SVM[J]. Mathematical Programming,2011,127(1):3-30.