基于遗传算法和BP神经网络的异纤图像识别方法研究
|
摘要
棉花中异性纤维的存在会严重影响棉纺织品的质量,从而影响棉纺织企业的经济效益。人工挑拣棉花异纤效率慢,漏检及错检率高。随着计算机视觉技术的发展,自动检测异性纤维技术得到了广泛的重视。本文对基于图像处理技术的异纤检测进行了研究,提出了一种基于遗传算法和BP神经网络的异纤识别算法。
首先,利用采用直方图均衡、中值滤波方法对异纤图像进行预处理,消除噪声影响。在分析各种分割算法的基础上,采用形态学方法对异纤图像进行分割。
然后,分析异纤图像的特点,选择具有类内距较小,类间距较大的特征。本文采用灰度共生矩阵方法计算图像共生矩阵熵、能量、惯性矩、相关4个纹理参数,分别计算各个参数的均值和方差,得到一幅图像的8个纹理特征;并选定六尺度、四方向的Gabor波滤波器组对图像进行滤波,得到24个图像,分别计算它们的均值和均方差,得到一幅图像的48维Gabor特征,并根据不同图像同类特征的方差,将48维降为36维。
最后,本文提出一种基于遗传算法的改进BP神经网络方法,对异纤图像进行识别分类。首先根据异纤图像特征的个数确定BP神经网络的输入层神经元个数;根据识别异纤的种类,确定3个输出层神经元,可识别8种类型的异纤。选定神经网络结构后,对随机给定的权值和阈值进行顺序编码,由神经网络的训练误差得出遗传算法的适应度函数,遗传算法对初始权值和阈值进行优化,得到优化后的BP神经网络。利用测试样本对BP神经网络进行训练,然后根据训练后的BP神经网络对测试样本进行识别。
实验结果表明,提出算法在异纤检测、识别上取得了较好的结果,具有较好的应用前景。
The profile fiber existing in cotton affects the quality of textiles, which will reduce the economic benefits of cotton textile enterprises. For conventional manual picking method, the efficiency is very low, and has the high probability of missed and wrong detection. With the development of computer vision, the technology of automatic detection has drawn widely attention. In this thesis, we research on the detection method of profile fiber based on image processing technology, and propose a detection method based on genetic algorithm and BP neural network.
Firstly, we apply histogram equalization and media filtering method into the fiber image to reduce the effect of noise. The effects of several segmentation methods are analyzed, and we select the morphological method as the segmentation algorithm in this thesis.
Secondly, we choose the feature with the small distance within the class and the larger distance between the class based on analyzing the character of fiber images. In this thesis, we adopt the gray-level co-occurrence matrix to depict the texture of fiber image because of this capability of blending spatial interaction with gray-level distribution. For the co-occurrence matrix, we calculate its entropy, energy, moment of inertia and relevant, and get a feature vector with 8-dimensions by calculating mean and variance of the four parameters. For the second feature, we select Gabor wavelet filter sets with six scales and four directions to filter the image to get 24 images, and get the feature with 48-dimensions by calculating the mean and variance of the 24 Gabor images. According to the calculating the variance of the same feature between different images, we reduce the 48-dimensions features into 36-dimensions.
In the end, we propose an improved BP neural network based on genetic algorithm, and adopt the method to classify the different kinds of the profile fiber. Firstly, we determine the number of neurons of BP neural network at input layer based on the dimension numbers of the feature. Then, based on the number of the kinds of profile fibers, we get the number of neurons at output layer. After selecting the neural network structure, we code the random weights and thresholds. And the improved BP neural network is achieved by optimizing the initial weights and thresholds using generetic algorithm. We train the improved neural network using training sets, and classify the test sets based on the trained neural network.
Experimental results show that the proposed method can efficiently detect and classify the profile fiber. And the proposed method has a good prospect.
首先,利用采用直方图均衡、中值滤波方法对异纤图像进行预处理,消除噪声影响。在分析各种分割算法的基础上,采用形态学方法对异纤图像进行分割。
然后,分析异纤图像的特点,选择具有类内距较小,类间距较大的特征。本文采用灰度共生矩阵方法计算图像共生矩阵熵、能量、惯性矩、相关4个纹理参数,分别计算各个参数的均值和方差,得到一幅图像的8个纹理特征;并选定六尺度、四方向的Gabor波滤波器组对图像进行滤波,得到24个图像,分别计算它们的均值和均方差,得到一幅图像的48维Gabor特征,并根据不同图像同类特征的方差,将48维降为36维。
最后,本文提出一种基于遗传算法的改进BP神经网络方法,对异纤图像进行识别分类。首先根据异纤图像特征的个数确定BP神经网络的输入层神经元个数;根据识别异纤的种类,确定3个输出层神经元,可识别8种类型的异纤。选定神经网络结构后,对随机给定的权值和阈值进行顺序编码,由神经网络的训练误差得出遗传算法的适应度函数,遗传算法对初始权值和阈值进行优化,得到优化后的BP神经网络。利用测试样本对BP神经网络进行训练,然后根据训练后的BP神经网络对测试样本进行识别。
实验结果表明,提出算法在异纤检测、识别上取得了较好的结果,具有较好的应用前景。
The profile fiber existing in cotton affects the quality of textiles, which will reduce the economic benefits of cotton textile enterprises. For conventional manual picking method, the efficiency is very low, and has the high probability of missed and wrong detection. With the development of computer vision, the technology of automatic detection has drawn widely attention. In this thesis, we research on the detection method of profile fiber based on image processing technology, and propose a detection method based on genetic algorithm and BP neural network.
Firstly, we apply histogram equalization and media filtering method into the fiber image to reduce the effect of noise. The effects of several segmentation methods are analyzed, and we select the morphological method as the segmentation algorithm in this thesis.
Secondly, we choose the feature with the small distance within the class and the larger distance between the class based on analyzing the character of fiber images. In this thesis, we adopt the gray-level co-occurrence matrix to depict the texture of fiber image because of this capability of blending spatial interaction with gray-level distribution. For the co-occurrence matrix, we calculate its entropy, energy, moment of inertia and relevant, and get a feature vector with 8-dimensions by calculating mean and variance of the four parameters. For the second feature, we select Gabor wavelet filter sets with six scales and four directions to filter the image to get 24 images, and get the feature with 48-dimensions by calculating the mean and variance of the 24 Gabor images. According to the calculating the variance of the same feature between different images, we reduce the 48-dimensions features into 36-dimensions.
In the end, we propose an improved BP neural network based on genetic algorithm, and adopt the method to classify the different kinds of the profile fiber. Firstly, we determine the number of neurons of BP neural network at input layer based on the dimension numbers of the feature. Then, based on the number of the kinds of profile fibers, we get the number of neurons at output layer. After selecting the neural network structure, we code the random weights and thresholds. And the improved BP neural network is achieved by optimizing the initial weights and thresholds using generetic algorithm. We train the improved neural network using training sets, and classify the test sets based on the trained neural network.
Experimental results show that the proposed method can efficiently detect and classify the profile fiber. And the proposed method has a good prospect.
引文
[1]韩培嵩,竺林.异性纤维溯源及对策[J].中国棉花加工,2000,(4) 20-21.
[2]陆振挺,陈玉峰,张新英.异纤清除技术的发展探讨.上海纺织科技[J].2008,36(1):9-11.
[3]吕守涛.异性纤维检测系统研究及硬件实现[D],北京工业大学,2004
[4]赵莹,高隽,汪荣贵等.一种新的广义最近邻方法研究[J].电子学报,2004,32(12):196-198.
[5]李彬,刘冀伟,韩鸿哲等.基于步态特征的快速身份识别方法[J].计算机工程与应用,2004(22),60.62.
[6] Manuele Bicego,Umberto Castellani,Vi~orio Murino. A hiddenmarkov model approach for appearance-based 3D object recognition[J].Patern Recongnition Letters,2005(26):2588-2599.
[7] Bo GP,Kyoung ML,Sang UL.Recognition of partially occluded objects using probabilistic ARG based matching[J].ComputVision Image Understanding,2003,3(90):217-241.
[8] Zheng feng Zhu,Ming Tang,Hanging Lu.A new robust circular Gabor based object matching by using weighted Hausdorf distance[J].Patern Recongnition Leters,2004(25):515-523.
[9] Tadashi Kondo,Abhijit S.Pand ya recognition of X-ray images by using revised GMDH type neural networks[J].Knowledge-Based Intelligent Information an d Engineering Systems,2003,2774:849-855.
[10] Zhihong Yao,Minrui Fei,Kang Li,et a1.Recognition of blue-green algae in lakes using distributive genetic algorithm-based neural networks[J].Neuro computing,2007,70(4):641-647.
[11]杨文柱,李道亮等,基于自动视觉检测的棉花异性纤维分类系统[J].农业机械学报,2 0 0 9,40(12),177―181.
[12]郭彩霞,袁建畅,金守峰.基于BP网络的棉花中异性纤维识别算法[J].西安工程科技学院学报,2006,20(5): 542-544.
[13]金守峰,张慧,冯涛.基于计算机视觉的棉花异性纤维检测识别算法[J].中国棉花加工2008(6):23-24.
[14]林宁,沙涛.基于数字图像处理的棉花异纤检测研究[J].电气电子报,2010,2(32):41-42.
[15]符宝鼎,袁建畅,郭彩霞基于RGB颜色模型棉花杂质识别算法[J].北京纺,2005,5(26):48-51.
[16]刘双喜,王金星,郑文秀.基于小波极值变换的棉花异性纤维边缘检测[J].农机化研究,2009 (8): 42-43/46.
[17]鲍义东,熊馨.基于Canny算子的棉花异性纤维检测算法的研究[J].牡丹江师范学院学报(自然科学版) ,2009(1):11-12.
[18]付腾,赵建军.基于空间模型的棉花异性纤维识别算法[J].郑州轻工业学院学报,2009.14(1): 87-90
[19]杨文柱,李道亮等,棉花异性纤维图像分割方法[J].农业机械学报,2009,40(3):156-160.
[20]郑文秀,刘双喜等.基于Mean-shift的棉花异性纤维图像分割[J].山东农业大学学报, 2009,40(2):224-228.
[21]周阳,朱邦太.一种基于视频技术的棉花异性纤维分拣方法[J].河南科技大学学报,2008,29(2):90-93.
[22]夏心怡,苏真伟,李国辉.基于形状特征的棉花异性纤维图像分割方法[J].农机化研究,2010,32(7):13-16.
[23]陈廷,张维龙.分形维数在原棉异纤图像处理中的应用[J].测试技术学报,2009, 20(3):248-252.
[24]杨文柱,李道亮等.基于光谱分析的棉花异性纤维最佳波段选择方法[J].农业工程学报, 2009,25(10): 186-191.
[25]徐飞,施晓红. MATLAB应用图像处理[M].西安:西安电子科技大学出版社,2002:140-162.
[26]冈萨雷斯.数字图像处理[M].北京:电子工业出版社,2003:61-174.
[27]朱锦雷.基于机器视觉的布匹表面质量快速检测算法的研究[D].武汉:湖北工业大学,2008:6-14.
[28] Otsu N.A threshold selection method from grey-level histograms[J].IEEE Trans System.Man cybernet,1979,SMC-9:62-66.
[29]王凤朝,黄树采,韩朝超.基于改进的二维Otsu法的图像分割法[J].航空计算技术,2008,38(4):4-7.
[30]龚炜,石青云.数字空间中的数学形态学理论及应用[M].北京:科学出版社,1997,150-151.
[31]王树文,闫成新.数学形态学在图像处理中的应用[J].计算机工程与应用,2004(32): 89-92.
[32] Horngw B,Peng Jianwen,Chen C Y. A New Image BasedReal Time Flame Detection Method Using Color Analysis[c] . Tucson,Arizona:Proc of IEEE International Confer - ence on Networking, Sensing and Control,IEEE Press,2005:100 -105.
[33] Milan Sonka,Vaclav Hlavac,Roger Boyle著艾海周武勃等议.图像处理、分析与机器视觉.人民邮电出版社,2003.
[34]李钰,孟祥萍.基于Gabor滤波器的图像纹理特征提取[J].长春工业大学学报,2008,29(1):78-81
[35] Jain A K, Farrokhnia F. Unsupervised texture segmentation using Gabor filters[J]. Pattern Recognition,1991,24(12):1167-1186.
[36] Polzleitner W. Defect detection on wooden surface using Gabor filters with evolutionary algorithm design [C].IEEE I-JCNN,2001.Washington:IEEE,2001: 750-755.
[37] Man junath B S,Ma W Y.Texture feature forbrowsing and retrieval of image data[J].IEEE Transaction on PAMI,1996,18(8):837-842.
[38]刘明霞,徐萍.基于Gabor变换的纹理分类与识别,电脑开发与应用,2008,21(4):2-3.
[39]肖鹏,徐军,陈少冲,纹理特征提取方法,电子科技,2010,23(6):49-51.
[40]李建萍,付丽琴,韩焱.用于特征提取的Gabor滤波器参数设计,光学与光电技术,2010,8(3):79-83.
[41]闻新,周露等Matlab神经网络仿真与应用[M],科学出版社,2003.
[42]杨晓帆,陈延槐.人工神经网络固有的优点和缺点[J];计算机科学;1994 21(2)
[43]武美先,张学良,温淑花.BP神经网络及其改进[J].太原科技大学学报,2005,26 (2):120-125.
[44]谭显胜,周铁军.BP算法改进方法的研究进展[J].怀化学院学报,2006,25(2):126一130.
[45]穆阿华,周绍磊等.利用遗传算法改进BP学习算法[J].计算机仿真,2005,2(2)150-151.
[46]蔡自兴,孙国荣,李枚毅.基于改进遗传算法的多示例神经网络优化[J].计算机应用,2005,25(10):2387-2389,2412.
[47]董恩国,张蕾.基于遗传神经网络算法的发动机性能预测[J].起重运输机械,2005(9):36-40
[48]李莉,基于遗传算法的BP神经网络在织物染色配色中的应用研究.青岛大学20070531
[49]高大启.有教师的线性基本函数前向三层神经网络结构研究[J].计算机学报,1998,21(1):80-86.
[2]陆振挺,陈玉峰,张新英.异纤清除技术的发展探讨.上海纺织科技[J].2008,36(1):9-11.
[3]吕守涛.异性纤维检测系统研究及硬件实现[D],北京工业大学,2004
[4]赵莹,高隽,汪荣贵等.一种新的广义最近邻方法研究[J].电子学报,2004,32(12):196-198.
[5]李彬,刘冀伟,韩鸿哲等.基于步态特征的快速身份识别方法[J].计算机工程与应用,2004(22),60.62.
[6] Manuele Bicego,Umberto Castellani,Vi~orio Murino. A hiddenmarkov model approach for appearance-based 3D object recognition[J].Patern Recongnition Letters,2005(26):2588-2599.
[7] Bo GP,Kyoung ML,Sang UL.Recognition of partially occluded objects using probabilistic ARG based matching[J].ComputVision Image Understanding,2003,3(90):217-241.
[8] Zheng feng Zhu,Ming Tang,Hanging Lu.A new robust circular Gabor based object matching by using weighted Hausdorf distance[J].Patern Recongnition Leters,2004(25):515-523.
[9] Tadashi Kondo,Abhijit S.Pand ya recognition of X-ray images by using revised GMDH type neural networks[J].Knowledge-Based Intelligent Information an d Engineering Systems,2003,2774:849-855.
[10] Zhihong Yao,Minrui Fei,Kang Li,et a1.Recognition of blue-green algae in lakes using distributive genetic algorithm-based neural networks[J].Neuro computing,2007,70(4):641-647.
[11]杨文柱,李道亮等,基于自动视觉检测的棉花异性纤维分类系统[J].农业机械学报,2 0 0 9,40(12),177―181.
[12]郭彩霞,袁建畅,金守峰.基于BP网络的棉花中异性纤维识别算法[J].西安工程科技学院学报,2006,20(5): 542-544.
[13]金守峰,张慧,冯涛.基于计算机视觉的棉花异性纤维检测识别算法[J].中国棉花加工2008(6):23-24.
[14]林宁,沙涛.基于数字图像处理的棉花异纤检测研究[J].电气电子报,2010,2(32):41-42.
[15]符宝鼎,袁建畅,郭彩霞基于RGB颜色模型棉花杂质识别算法[J].北京纺,2005,5(26):48-51.
[16]刘双喜,王金星,郑文秀.基于小波极值变换的棉花异性纤维边缘检测[J].农机化研究,2009 (8): 42-43/46.
[17]鲍义东,熊馨.基于Canny算子的棉花异性纤维检测算法的研究[J].牡丹江师范学院学报(自然科学版) ,2009(1):11-12.
[18]付腾,赵建军.基于空间模型的棉花异性纤维识别算法[J].郑州轻工业学院学报,2009.14(1): 87-90
[19]杨文柱,李道亮等,棉花异性纤维图像分割方法[J].农业机械学报,2009,40(3):156-160.
[20]郑文秀,刘双喜等.基于Mean-shift的棉花异性纤维图像分割[J].山东农业大学学报, 2009,40(2):224-228.
[21]周阳,朱邦太.一种基于视频技术的棉花异性纤维分拣方法[J].河南科技大学学报,2008,29(2):90-93.
[22]夏心怡,苏真伟,李国辉.基于形状特征的棉花异性纤维图像分割方法[J].农机化研究,2010,32(7):13-16.
[23]陈廷,张维龙.分形维数在原棉异纤图像处理中的应用[J].测试技术学报,2009, 20(3):248-252.
[24]杨文柱,李道亮等.基于光谱分析的棉花异性纤维最佳波段选择方法[J].农业工程学报, 2009,25(10): 186-191.
[25]徐飞,施晓红. MATLAB应用图像处理[M].西安:西安电子科技大学出版社,2002:140-162.
[26]冈萨雷斯.数字图像处理[M].北京:电子工业出版社,2003:61-174.
[27]朱锦雷.基于机器视觉的布匹表面质量快速检测算法的研究[D].武汉:湖北工业大学,2008:6-14.
[28] Otsu N.A threshold selection method from grey-level histograms[J].IEEE Trans System.Man cybernet,1979,SMC-9:62-66.
[29]王凤朝,黄树采,韩朝超.基于改进的二维Otsu法的图像分割法[J].航空计算技术,2008,38(4):4-7.
[30]龚炜,石青云.数字空间中的数学形态学理论及应用[M].北京:科学出版社,1997,150-151.
[31]王树文,闫成新.数学形态学在图像处理中的应用[J].计算机工程与应用,2004(32): 89-92.
[32] Horngw B,Peng Jianwen,Chen C Y. A New Image BasedReal Time Flame Detection Method Using Color Analysis[c] . Tucson,Arizona:Proc of IEEE International Confer - ence on Networking, Sensing and Control,IEEE Press,2005:100 -105.
[33] Milan Sonka,Vaclav Hlavac,Roger Boyle著艾海周武勃等议.图像处理、分析与机器视觉.人民邮电出版社,2003.
[34]李钰,孟祥萍.基于Gabor滤波器的图像纹理特征提取[J].长春工业大学学报,2008,29(1):78-81
[35] Jain A K, Farrokhnia F. Unsupervised texture segmentation using Gabor filters[J]. Pattern Recognition,1991,24(12):1167-1186.
[36] Polzleitner W. Defect detection on wooden surface using Gabor filters with evolutionary algorithm design [C].IEEE I-JCNN,2001.Washington:IEEE,2001: 750-755.
[37] Man junath B S,Ma W Y.Texture feature forbrowsing and retrieval of image data[J].IEEE Transaction on PAMI,1996,18(8):837-842.
[38]刘明霞,徐萍.基于Gabor变换的纹理分类与识别,电脑开发与应用,2008,21(4):2-3.
[39]肖鹏,徐军,陈少冲,纹理特征提取方法,电子科技,2010,23(6):49-51.
[40]李建萍,付丽琴,韩焱.用于特征提取的Gabor滤波器参数设计,光学与光电技术,2010,8(3):79-83.
[41]闻新,周露等Matlab神经网络仿真与应用[M],科学出版社,2003.
[42]杨晓帆,陈延槐.人工神经网络固有的优点和缺点[J];计算机科学;1994 21(2)
[43]武美先,张学良,温淑花.BP神经网络及其改进[J].太原科技大学学报,2005,26 (2):120-125.
[44]谭显胜,周铁军.BP算法改进方法的研究进展[J].怀化学院学报,2006,25(2):126一130.
[45]穆阿华,周绍磊等.利用遗传算法改进BP学习算法[J].计算机仿真,2005,2(2)150-151.
[46]蔡自兴,孙国荣,李枚毅.基于改进遗传算法的多示例神经网络优化[J].计算机应用,2005,25(10):2387-2389,2412.
[47]董恩国,张蕾.基于遗传神经网络算法的发动机性能预测[J].起重运输机械,2005(9):36-40
[48]李莉,基于遗传算法的BP神经网络在织物染色配色中的应用研究.青岛大学20070531
[49]高大启.有教师的线性基本函数前向三层神经网络结构研究[J].计算机学报,1998,21(1):80-86.