基于深度学习的海底观测视频中鱼类的识别方法
|
摘要
针对海底环境恶劣、海底观测视频品质差导致视频中的海洋鱼类识别难的问题,提出一种基于深度学习的海洋鱼类识别方法。首先,将海底观测视频分解为图片,由于海底观测视频中存在较大比例的空白数据,使用背景差分法过滤不包含鱼类的图片,缩短处理全部数据的时间;然后,考虑到海底拍摄环境亮度低、场景模糊的实际情况,对图片基于暗通道先验算法进行预处理提高品质;最后,以卷积神经网络(CNN)为基础构建深度学习模型,并且提出了权重化特征的卷积过程,提高模型的鲁棒性。实验结果表明:面对较差品质的海底观测视频图片,在深度学习模型结构相同的条件下,与普通卷积神经网络模型相比,使用权重化卷积作为隐层并且加入预处理过程后,对海洋鱼类识别准确率的提升幅度达到23%,有助于实现对海底观测视频图片中海洋鱼类的精准识别。
As it is hard to recognize marine fishes occurred in submarine observation videos due to the bad undersea environment and low quality of the video,a recognition method based on deep learning was proposed.Firstly,the video was split into pictures,and as this type of video contains a large proportion of useless data,a background subtraction algorithm was used to filter the pictures without fish to save the time of processing all data.Then,considering the undersea environment is blurring with low bright,based on the dark channel prior algorithm,the pictures were preprocessed to improve their quality before recognition.Finally,a recognition deep learning model based on Convolutional Neural Network(CNN)was consructed with weighted convolution process to improve the robustness of the model.The experimental results show that,facing submarine observation video frames with poor quality,compared with traditional CNN,the method with preprocessing and weighted convolution as hidden layer can increase the recognition accuracy by 23%,contributing to the recognition of marine fishes in submarine observation video.
As it is hard to recognize marine fishes occurred in submarine observation videos due to the bad undersea environment and low quality of the video,a recognition method based on deep learning was proposed.Firstly,the video was split into pictures,and as this type of video contains a large proportion of useless data,a background subtraction algorithm was used to filter the pictures without fish to save the time of processing all data.Then,considering the undersea environment is blurring with low bright,based on the dark channel prior algorithm,the pictures were preprocessed to improve their quality before recognition.Finally,a recognition deep learning model based on Convolutional Neural Network(CNN)was consructed with weighted convolution process to improve the robustness of the model.The experimental results show that,facing submarine observation video frames with poor quality,compared with traditional CNN,the method with preprocessing and weighted convolution as hidden layer can increase the recognition accuracy by 23%,contributing to the recognition of marine fishes in submarine observation video.
引文
[1]李继龙,曹坤,丁放,等.基于渔获物统计的中国近海鱼类营养级结构变换及其与捕捞作业的关系[J].中国水产科学,2017,24(1):109-119.(LI J L,CAO K,DING F,et al.Changes in trophic-level structure of the main fish species caught by China and their relationship with fishing method[J].Journal of Fishery Sciences of China,2017,24(1):109-119.)
[2]岳冬冬,王鲁民,张勋,等.我国海洋捕捞装备与技术发展趋势研究[J].中国农业科技导报,2013,15(6):20-26.(YUE D D,WANG L M,ZHANG X,et al.The development trends of marine fishing equipment and technology in China[J].Journal of Agricultural Science and Technology,2013,15(6):20-26.)
[3]许枫,张乔,张纯,等.Walsh变换对鱼类特征识别的研究[J].应用声学,2015,34(5):465-470.(XU F,ZHANG Q,ZHANG C,et al.Walsh transform for fish identification[J].Applied Acoustics,2015,34(5):465-470.)
[4]张志强,牛智有,赵思明.基于机器视觉技术的淡水鱼品种识别[J].农业工程学报,2011,27(11):388-392.(ZHANG Z Q,NIU Z Y,ZHAO S M.Identification of freshwater fish species based on computer vision[J].Transactions of the Chinese Society of Agricultural Engineering,2011,27(11):388-392.)
[5]姚润璐,桂詠雯,黄秋桂.基于机器视觉的淡水鱼品种识别[J].微型机与应用,2017,36(24):37-39.(YAO R L,GUI Y W,HUANG Q G.Recognition of freshwater fish species based on machine vision[J].Microcomputer and Applications,2017,36(24):37-39.)
[6]林明旺.深度学习在鱼类图像识别与分类中的应用[J].数字技术与应用,2017,63(4):96-97.(LIN M W.Application of deep learning in fish image recognition and classification[J].Digital Technology and Application,2017,63(4):96-97.)
[7]沈瑜,王新新.基于背景减法和帧间差分法的视频运动目标检测方法[J].自动化与仪器仪表,2017(4):122-124.(SHENY,WANG X X.Video moving target detection method based on background subtraction and interframe difference method[J].Automation and Instrumentation,2017(4):122-124.)
[8]蒋明敏.基于FPGA的LCD伽马校正研究[D].南京:南京林业大学,2016:25-27.(JIANG M M.Research on LCD Gamma correction based on FPGA[D].Nanjing:Nanjing Forestry University,2016:25-28.)
[9]HE K,SUN J,TANG X.Single image haze removal using dark channel prior[J].IEEE transactions on Pattern Analysis and Machine Intelligence,2011,33(12):2341-2353.
[10]张波,唐启升,金显仕.东海高营养层次鱼类功能群及其主要种类[J].中国水产科学,2007,14(6):939-949.(ZHANG B,TANG Q S,JIN X S.Functional groups of fish assemblages and their major species at high trophic level in the East China Sea[J].Journal of Fishery Sciences of China,2007,14(6):939-949.)
[11]宋超,侯俊利,赵峰,等.春、秋季东海大桥海上风电场水域鱼类群落结构[J].海洋科学,2017,41(6):34-40.(SONG C,HOU J L,ZHAO F,et al.Fish community structure in the offshore wind farm of Donghai Bridge in spring and autumn[J].O-cean Science,2017,41(6):34-40.)
[12]RAWAT W,WANG Z.Deep convolutional neural networks for image classification:a comprehensive review[J].Neural Computation,2017,29(9):2352-2449.
[13]SUTSKEVER I,MARTENS J,DAHL G,et al.On the importance of initialization and momentum in deep learning[C]//Proceedings of the 30th International Conference on Machine Learning:Vol.28.Atlanta,GA:JMLR,2013:Ⅲ-1139-Ⅲ-1147.
[14]ABADI M,BARHAM P,CHEN J,et al.TensorFlow:a system for large-scale machine learning[C]//Proceedings of the 12th USENIX Conference on Operating Systems Design and Implementation.Berkeley,CA:USENIX Association,2016:265-283.
[15]李彦冬,郝宗波,雷航.卷积神经网络研究综述[J].计算机应用,2016,36(9):2508-2515.(LI Y D,HAO Z B,LEI H.Survey of convolutional neural network[J].Journal of Computer Applications,2016,36(9):2508-2515.)
[16]DENG J,DONG W,SOCHER R,et al.Image Net:a large-scale hierarchical image database[C]//Proceedings of the 2009 IEEEConference on Computer Vision and Pattern Recognition.Washington,DC:IEEE Computer Society,2009:248-255.
[2]岳冬冬,王鲁民,张勋,等.我国海洋捕捞装备与技术发展趋势研究[J].中国农业科技导报,2013,15(6):20-26.(YUE D D,WANG L M,ZHANG X,et al.The development trends of marine fishing equipment and technology in China[J].Journal of Agricultural Science and Technology,2013,15(6):20-26.)
[3]许枫,张乔,张纯,等.Walsh变换对鱼类特征识别的研究[J].应用声学,2015,34(5):465-470.(XU F,ZHANG Q,ZHANG C,et al.Walsh transform for fish identification[J].Applied Acoustics,2015,34(5):465-470.)
[4]张志强,牛智有,赵思明.基于机器视觉技术的淡水鱼品种识别[J].农业工程学报,2011,27(11):388-392.(ZHANG Z Q,NIU Z Y,ZHAO S M.Identification of freshwater fish species based on computer vision[J].Transactions of the Chinese Society of Agricultural Engineering,2011,27(11):388-392.)
[5]姚润璐,桂詠雯,黄秋桂.基于机器视觉的淡水鱼品种识别[J].微型机与应用,2017,36(24):37-39.(YAO R L,GUI Y W,HUANG Q G.Recognition of freshwater fish species based on machine vision[J].Microcomputer and Applications,2017,36(24):37-39.)
[6]林明旺.深度学习在鱼类图像识别与分类中的应用[J].数字技术与应用,2017,63(4):96-97.(LIN M W.Application of deep learning in fish image recognition and classification[J].Digital Technology and Application,2017,63(4):96-97.)
[7]沈瑜,王新新.基于背景减法和帧间差分法的视频运动目标检测方法[J].自动化与仪器仪表,2017(4):122-124.(SHENY,WANG X X.Video moving target detection method based on background subtraction and interframe difference method[J].Automation and Instrumentation,2017(4):122-124.)
[8]蒋明敏.基于FPGA的LCD伽马校正研究[D].南京:南京林业大学,2016:25-27.(JIANG M M.Research on LCD Gamma correction based on FPGA[D].Nanjing:Nanjing Forestry University,2016:25-28.)
[9]HE K,SUN J,TANG X.Single image haze removal using dark channel prior[J].IEEE transactions on Pattern Analysis and Machine Intelligence,2011,33(12):2341-2353.
[10]张波,唐启升,金显仕.东海高营养层次鱼类功能群及其主要种类[J].中国水产科学,2007,14(6):939-949.(ZHANG B,TANG Q S,JIN X S.Functional groups of fish assemblages and their major species at high trophic level in the East China Sea[J].Journal of Fishery Sciences of China,2007,14(6):939-949.)
[11]宋超,侯俊利,赵峰,等.春、秋季东海大桥海上风电场水域鱼类群落结构[J].海洋科学,2017,41(6):34-40.(SONG C,HOU J L,ZHAO F,et al.Fish community structure in the offshore wind farm of Donghai Bridge in spring and autumn[J].O-cean Science,2017,41(6):34-40.)
[12]RAWAT W,WANG Z.Deep convolutional neural networks for image classification:a comprehensive review[J].Neural Computation,2017,29(9):2352-2449.
[13]SUTSKEVER I,MARTENS J,DAHL G,et al.On the importance of initialization and momentum in deep learning[C]//Proceedings of the 30th International Conference on Machine Learning:Vol.28.Atlanta,GA:JMLR,2013:Ⅲ-1139-Ⅲ-1147.
[14]ABADI M,BARHAM P,CHEN J,et al.TensorFlow:a system for large-scale machine learning[C]//Proceedings of the 12th USENIX Conference on Operating Systems Design and Implementation.Berkeley,CA:USENIX Association,2016:265-283.
[15]李彦冬,郝宗波,雷航.卷积神经网络研究综述[J].计算机应用,2016,36(9):2508-2515.(LI Y D,HAO Z B,LEI H.Survey of convolutional neural network[J].Journal of Computer Applications,2016,36(9):2508-2515.)
[16]DENG J,DONG W,SOCHER R,et al.Image Net:a large-scale hierarchical image database[C]//Proceedings of the 2009 IEEEConference on Computer Vision and Pattern Recognition.Washington,DC:IEEE Computer Society,2009:248-255.