基于计算机视觉和GA-SVM的梭子蟹体重预测
|
摘要
以梭子蟹为研究对象,利用计算机视觉技术对其进行测量.通过CCD相机获取不同生长情况下的梭子蟹图像,采用图像处理技术对图像进行分割处理,计算获得的投影面积、全甲宽与甲长参数;利用图像获取的尺寸参数对梭子蟹体重进行预测,发现梭子蟹投影面积、全甲宽、甲长与体重具有正相关性;并采用遗传优化(GA)的支持向量机(SVM)建立梭子蟹体重回归预测模型.实测结果表明,梭子蟹体重预测值与实测值平均绝对百分比误差(MAPE)为2.23%,均方根误差(RMSE)为5.80 g,优于BP神经网络和参数未优化的SVM预测.证明基于计算机视觉与遗传优化支持向量机(GA-SVM)的梭子蟹体重预测方法能够达到梭子蟹体重测量要求.
In the paper, computer vision technology is applied to make non-human-interference measurement of swimming crabs. First, the images of swimming crabs under different development stages are obtained by CCD camera. Then, the projected area of the swimming crab can be computed based on the image, so are the full carapace width and carapace length. By examining the positive correlation among the projected area, full carapace width, carapace length and weight of crab, the weight can hence be predicted. The weight prediction model of swimming crab based on Genetic Algorithm(GA) and Support Vector Machine(SVM) is established. The results suggest that the Mean Absolute Percent Error(MAPE) of body weight reads 2.23%, and the Mean Square Error(MSE) is found to be 5.80 g. Compared with BP neural network and SVM, this method shows a better forecasting precision. The experimental results indicate that the method based on computer vision and Genetic Algorithm-Support Vector Machine(GA-SVM) can better predict the body weight of the swimming crabs.
In the paper, computer vision technology is applied to make non-human-interference measurement of swimming crabs. First, the images of swimming crabs under different development stages are obtained by CCD camera. Then, the projected area of the swimming crab can be computed based on the image, so are the full carapace width and carapace length. By examining the positive correlation among the projected area, full carapace width, carapace length and weight of crab, the weight can hence be predicted. The weight prediction model of swimming crab based on Genetic Algorithm(GA) and Support Vector Machine(SVM) is established. The results suggest that the Mean Absolute Percent Error(MAPE) of body weight reads 2.23%, and the Mean Square Error(MSE) is found to be 5.80 g. Compared with BP neural network and SVM, this method shows a better forecasting precision. The experimental results indicate that the method based on computer vision and Genetic Algorithm-Support Vector Machine(GA-SVM) can better predict the body weight of the swimming crabs.
引文
[1]穆占昆,杨振国,周玉,等.中国对虾和三疣梭子蟹混养试验[J].水产科学,2001,20(5):16-18.
[2]王春琳,母昌考,李荣华,等.三疣梭子蟹单体筐养高产高效生产技术[J].中国水产,2013,1(1):72-76.
[3]李长缨,滕光辉,赵春江,等.利用计算机视觉技术实现对温室植物生长的无损监测[J].农业工程学报,2003,19(3):140-143.
[4]徐建瑜,崔绍荣,苗香雯,等.计算机视觉技术在水产养殖中的应用与展望[J].农业工程学报,2005,21(8):174-178.
[5]Lee D J,Archibald J K,Schoenberger R B,et al.Contour matching for fish species recognition and migration monitoring[M]//Smolinski T G,Milanova M G,Hassanien A E.Applications of Computational Intelligence in Biology.Heidelberg:Springer,2008:183-207.
[6]Zion B,Alchanatis V,Ostrovsky V,et al.Classification of guppies’(Poecilia reticulata)gender by computer vision[J].Aquacultural Engineering,2008,38(2):97-104.
[7]Lines J A,Tillett R D,Ross L G,et al.An automatic image-based system for estimating the mass of free-swimming fish[J].Computers and Electronics in Agriculture,2001,31(2):151-168.
[8]Liang Y T,Chiou Y C.Machine vision-based automatic raw fish handling and weighing system of Taiwan Tilapia[M]//Chien B C,Hong T P,Chen S M,et al.Nextgeneration Applied Intelligence.Berlin:Springer,2009:711-720.
[9]刘伟,谭鹤群,黄丹,等.白鲢质量与截面积沿体长方向分布模型[J].农业工程学报,2012,28(12):288-292.
[10]张超,徐建瑜,王文静.基于机器视觉的梭子蟹质量估计方法研究[J].宁波大学学报(理工版),2014,27(2):49-51.
[11]魏崇德,陈永寿.浙江动物志甲壳类[M].杭州:浙江科学技术出版社,1991:356.
[12]Rother C,Kolmogorov V,Blake A.“Grab Cut”:Interactive foreground extraction using iterated graph cuts[J].ACM Transactions on Graphics,2004,23(3):309-314.
[13]伊力哈木·亚尔买买提.基于改进的自适应分水岭图像分割方法研究[J].计算机仿真,2013,30(2):373-377.
[14]方瑞明.支持向量机理论及其应用分析[M].北京:中国电力出版社,2007.
[15]周辉仁,郑丕谔,赵春秀.基于遗传算法的LS-SVM参数优选及其在经济预测中的应用[J].计算机应用,2007,27(6):1418-1419.
[16]G?nen M,Alpaydin E.Localized multiple kernel learning[C].Proceedings of the 25th International Conference on Machine Learning,ACM,2008:352-359.
[17]雷英杰,张善文.MATLAB遗传算法工具箱及应用[M].西安:西安电子科技大学出版社,2014.
[18]张婉琳.遗传算法优化支持向量机的交通流量预测[J].激光杂志,2014(12):116-119.
[2]王春琳,母昌考,李荣华,等.三疣梭子蟹单体筐养高产高效生产技术[J].中国水产,2013,1(1):72-76.
[3]李长缨,滕光辉,赵春江,等.利用计算机视觉技术实现对温室植物生长的无损监测[J].农业工程学报,2003,19(3):140-143.
[4]徐建瑜,崔绍荣,苗香雯,等.计算机视觉技术在水产养殖中的应用与展望[J].农业工程学报,2005,21(8):174-178.
[5]Lee D J,Archibald J K,Schoenberger R B,et al.Contour matching for fish species recognition and migration monitoring[M]//Smolinski T G,Milanova M G,Hassanien A E.Applications of Computational Intelligence in Biology.Heidelberg:Springer,2008:183-207.
[6]Zion B,Alchanatis V,Ostrovsky V,et al.Classification of guppies’(Poecilia reticulata)gender by computer vision[J].Aquacultural Engineering,2008,38(2):97-104.
[7]Lines J A,Tillett R D,Ross L G,et al.An automatic image-based system for estimating the mass of free-swimming fish[J].Computers and Electronics in Agriculture,2001,31(2):151-168.
[8]Liang Y T,Chiou Y C.Machine vision-based automatic raw fish handling and weighing system of Taiwan Tilapia[M]//Chien B C,Hong T P,Chen S M,et al.Nextgeneration Applied Intelligence.Berlin:Springer,2009:711-720.
[9]刘伟,谭鹤群,黄丹,等.白鲢质量与截面积沿体长方向分布模型[J].农业工程学报,2012,28(12):288-292.
[10]张超,徐建瑜,王文静.基于机器视觉的梭子蟹质量估计方法研究[J].宁波大学学报(理工版),2014,27(2):49-51.
[11]魏崇德,陈永寿.浙江动物志甲壳类[M].杭州:浙江科学技术出版社,1991:356.
[12]Rother C,Kolmogorov V,Blake A.“Grab Cut”:Interactive foreground extraction using iterated graph cuts[J].ACM Transactions on Graphics,2004,23(3):309-314.
[13]伊力哈木·亚尔买买提.基于改进的自适应分水岭图像分割方法研究[J].计算机仿真,2013,30(2):373-377.
[14]方瑞明.支持向量机理论及其应用分析[M].北京:中国电力出版社,2007.
[15]周辉仁,郑丕谔,赵春秀.基于遗传算法的LS-SVM参数优选及其在经济预测中的应用[J].计算机应用,2007,27(6):1418-1419.
[16]G?nen M,Alpaydin E.Localized multiple kernel learning[C].Proceedings of the 25th International Conference on Machine Learning,ACM,2008:352-359.
[17]雷英杰,张善文.MATLAB遗传算法工具箱及应用[M].西安:西安电子科技大学出版社,2014.
[18]张婉琳.遗传算法优化支持向量机的交通流量预测[J].激光杂志,2014(12):116-119.