局部二值模式的改进及其在工业X射线图像中的应用研究
|
摘要
X射线成像技术，如断层成像技术（CT, Computed Tomography）和数字辐射成像技术（DR, Digital Radiography），可以通过图像无损地展示被测物体的内部结构，广泛用于工业探伤和医学检测。在工业应用中，X射线无损检测尤其是实时批量检测有一定的时间限制，这就要求相应的图像处理方法能够快速完成检测任务；在DR图和CT体数据中，铸件的灰度对比度和缺陷的分布呈现出一定的局部特性，这就要求相应的图像处理方法能够很好地提取局部特征，能够得到较为准确的边缘检测或特征提取结果。但是传统的图像检测方法受图像数据尺寸或者图像模糊的影响，难以在这两个方面上取得平衡。
为了提高检测效率，本论文将纹理描述算子局部二值模式（LBP，Local BinaryPattern）用于X射线图像的图像增强、边缘提取和特征提取等图像处理中。LBP通过加权代表圆域内像素间灰度差异的二值数值来描述纹理信息，具有计算方式简单性、信息提取局部性和单调灰度不变性等优势。本论文根据DR图像和CT体数据的特点，在保持LBP这些优势的同时，从加权方式、待处理数据特点和比较函数上对LBP作相应的改进，提高了LBP对方向和空间信息的提取能力，增强了LBP对边缘的捕捉能力，弥补了LBP对灰度差异程度描述能力的不足。
铸件DR图像中缺陷区域的灰度对比度不高，利用LBP的信息提取局部化优势，可以增强含缺陷的DR图，但LBP无法同时增强含有不同方向缺陷的子图。针对LBP对方向信息提取不充分的缺点，我们改进了LBP的加权过程，利用有限线积分变换(FLIT, Finite Line Integral Transform)的积分值来确定图像主方向，按主方向选定合适的权重排列方式来加权对应的二值数值，获得含有方向信息的LBP值。该利用FLIT方向信息改进LBP的算法称为FLIT-LBP。因FLIT的方向优势，FLIT-LBP算法可以增强不同方向的缺陷；因LBP信息提取局部化的优势以及单调灰度不变性，FLIT-LBP算法可以增强不同对比度下的缺陷。特别的，FLIT-LBP可以对强灰度对比度的缺陷进行增强的同时，增强同一区域内灰度对比度相对偏弱的缺陷。经FLIT-LBP增强后，缺陷的特征更为突出，在此基础上提取相应的纹理特征，可以在一定程度上提高识别背景、铸件强边缘和缺陷的识别率。
CT体数据尺寸较大，利用LBP的计算方式简单性可以较为快速地获得边缘，但LBP对微小的灰度变化过于敏感。为了克服这个缺点，我们研究了一种融合FLIT和LBP空间信息的三维边缘提取算法FL-fusion。通过融合多空间方向的FLIT值来获得二值体数据，以满足LBP在提取边缘时对数据的分段常数化要求。通过融合多平面LBP值来获得三维边缘，提高LBP对空间信息尤其是三维边缘信息的捕捉能力。FL-fusion可以克服噪声影响，较快地提取细化的、连续的、封闭的、较为准确的三维边缘。值得注意的是，FL-fusion能够保留边缘中的细节信息，如裂纹的尖端，为特殊部位检测分析奠定了基础。该方法同样也适用于其他含有复杂结构的CT体数据边缘提取。
DR图像较为模糊，利用LBP对非单调灰度变化敏感的优势可以提取边缘，但LBP无法描述像素间灰度差异的程度，不能区分冗余的微小灰度变化和所要保留的较大的灰度变化。本论文针对此问题，利用相对光滑的比较函数来改进LBP，称为H-LBP。该算法通过嵌入含有单调递减性质的相对光滑H函数来逼近原来阶跃的比较函数S，并考虑了圆域内中心点和圆周邻点的灰度相似距离，以有区别地对待圆域内的灰度信息。另外，在该算法中增加了一个计数策略，以淘汰冗余的微小灰度变化。H-LBP计算简单，可以较为快速地完成边缘的提取；保留了对较大灰度变化的敏感性，增强了对灰度差异程度的描述能力，可以有区别地提取铸件边缘、灰度对比度较弱的小缺陷边缘和灰度不均的铸件号边缘，克服了噪声、伪影和图像模糊。H-LBP也可用于CT图像、背景复杂的图像和光照不均的图像。
本论文针对X射线图像特点和实际检测要求，对LBP作出了相应的改进。实际实验证实这些方法具有处理时间短和检测效果好两重优势，可有效地增强图像、提取边缘和识别缺陷，对工业X射线无损检测具有指导意义。
X-ray imaging techniques such as computed tomography (CT) and digitalradiography (DR), are widely been used on industrial inspetion and medical inprobing,as they are capable of showing the inner structure of tested object via images withoutdestructing object. In industrial applications, there are some time requests onnon-destructive detection by X-ray, especially on realtime detection, which ask for a fastdetection method. In DR image and CT volume data, the grayscale contrast ratios ofcastings and the distribution of defects are presented with local properties, which ask fora local information extraction method. However, traditional detecting methods are toughto keep the balance between processing time and final vision, due to the blurry and largesize of images.
In order to improve the efficiency of detection, local binary pattern (LBP), atexture discribtor, is exploted in X-ray images’ relative procession such as imageenhancement, edge extraction and feature extraction. LBP that has properties ascalculation simplicity, information locality, and monotic grayscale invariance, describesthe texture by calaulating the weighted sum of binary codes indicating the grayscaledifference in circular neighborhood. Acording to the traits of DR images and CT volumedata, and to profit from these advantages, LBP is improved respecting to aspects of theway to assign weights, the data to be handled, and the function to compare by. Suchimprovements has heightened the extraction capability of direction information andspatial information, improved the capture capability of edge information, andsuperinduced the description capability to the difference degree of grayscale.
Defects in castiings’ DR images are not very clear due to the low grayscale contrastratio. By utilizing the information locality, LBP can be used to enhance DR sub-imagewith defects, however, it cannot enhance DR sub-image with defects appearing indifferent directions. To overcome this shortcoming, the weighted sum calculation ofLBP is improved by using finite line integral transform (FLIT) to determine the maindirection of sub-image then to choose the right weight arrangement for the sumcalculation in LBP. The improved method is so-called FLIT-LBP. Involving thedirection property of FLIT, FLIT-LBP can enhance defects in different directions;Reserving the local advantages and grayscale invariance of LBP, FLIT-LBP can enhancedefects under different grayscale contrast ratios, especially defects in low contrast. Inaddition, after enhancement of FLIT-LBP, texture features of defects are emphasized so that relative texures can achieve a better recognition rate.
As CT volume data are conmmonly of large size, LBP is used to extract3D edgeby taking advantage of its calculation simplicity. Nevertheless, LBP is too sensitive tomicaro-changes in grayscale. In this case, we have studied a3D edge extraction methodcalled FL-fusion for CT volume data, by fusing spatial information of FLIT and LBP.Multiple direction information obtained by FLIT is fused to get binarizate volume data,so to satisfy the piecewise constant request of data in hand. Then multiple planeinformation obtained by LBP is fused to extract3D edges. FL-fusion method can obtainthin, continuous, occlusive and accurate3D edges without being affected by noises andartifacts, in a short time. A sifnifficant point is that FL-fusion also can maintain themicro feature of edges such as the tine of crack, which provides a base stone forparticular compoment analysis. This method can be used on kinds of CT volume data,whenever the inner structure simple or complicate.
Since DR images are blurry and large, LBP can be used to extract the edge due toits sensitivity to non-monotic grayscale changes. However, unable to describe thedegree of grayscale difference, LBP cannot distinguish randontant micro-changes anduseful changes of grayscale. To settle this problem, LBP is improved with a smoothcompare fuction. The so-called H-LBP embeds a comparative smooth and mononicdecreasing H funcrion into LBP to approaching original stepwise S function, andconsiders a similarity distance between surrounding point and centric point to describegrayscale relationships distinctively. Besides, a counting scheme is utilized to eliminateredundant micro-changes. With such modifications, H-LBP keeps the calculationsimplicity and maintains the sensitivity to larger grayscale changes while improves thedescription capability of grayscale difference degree. Thus, H-LBP is able to extractintegrated edges in blurry DR images, including small defect edges under low contrastratio and batch number edges under grayscale inhomogeneous conditions. Ourexperiments also show that H-LBP is applicable for CT images, complicate backgroundimages and illuminate inhomogeneous images.
To sum up, LBP is improved according to the traits of X-ray images and therequests of detection. These improved methods have advantages in both processing timeand detecting vision, feasible for image enhancement, edge extraction or defectclassification. Thus, our work plays a guiding role in indrustrial X-ray nondestructivedetection.
为了提高检测效率，本论文将纹理描述算子局部二值模式（LBP，Local BinaryPattern）用于X射线图像的图像增强、边缘提取和特征提取等图像处理中。LBP通过加权代表圆域内像素间灰度差异的二值数值来描述纹理信息，具有计算方式简单性、信息提取局部性和单调灰度不变性等优势。本论文根据DR图像和CT体数据的特点，在保持LBP这些优势的同时，从加权方式、待处理数据特点和比较函数上对LBP作相应的改进，提高了LBP对方向和空间信息的提取能力，增强了LBP对边缘的捕捉能力，弥补了LBP对灰度差异程度描述能力的不足。
铸件DR图像中缺陷区域的灰度对比度不高，利用LBP的信息提取局部化优势，可以增强含缺陷的DR图，但LBP无法同时增强含有不同方向缺陷的子图。针对LBP对方向信息提取不充分的缺点，我们改进了LBP的加权过程，利用有限线积分变换(FLIT, Finite Line Integral Transform)的积分值来确定图像主方向，按主方向选定合适的权重排列方式来加权对应的二值数值，获得含有方向信息的LBP值。该利用FLIT方向信息改进LBP的算法称为FLIT-LBP。因FLIT的方向优势，FLIT-LBP算法可以增强不同方向的缺陷；因LBP信息提取局部化的优势以及单调灰度不变性，FLIT-LBP算法可以增强不同对比度下的缺陷。特别的，FLIT-LBP可以对强灰度对比度的缺陷进行增强的同时，增强同一区域内灰度对比度相对偏弱的缺陷。经FLIT-LBP增强后，缺陷的特征更为突出，在此基础上提取相应的纹理特征，可以在一定程度上提高识别背景、铸件强边缘和缺陷的识别率。
CT体数据尺寸较大，利用LBP的计算方式简单性可以较为快速地获得边缘，但LBP对微小的灰度变化过于敏感。为了克服这个缺点，我们研究了一种融合FLIT和LBP空间信息的三维边缘提取算法FL-fusion。通过融合多空间方向的FLIT值来获得二值体数据，以满足LBP在提取边缘时对数据的分段常数化要求。通过融合多平面LBP值来获得三维边缘，提高LBP对空间信息尤其是三维边缘信息的捕捉能力。FL-fusion可以克服噪声影响，较快地提取细化的、连续的、封闭的、较为准确的三维边缘。值得注意的是，FL-fusion能够保留边缘中的细节信息，如裂纹的尖端，为特殊部位检测分析奠定了基础。该方法同样也适用于其他含有复杂结构的CT体数据边缘提取。
DR图像较为模糊，利用LBP对非单调灰度变化敏感的优势可以提取边缘，但LBP无法描述像素间灰度差异的程度，不能区分冗余的微小灰度变化和所要保留的较大的灰度变化。本论文针对此问题，利用相对光滑的比较函数来改进LBP，称为H-LBP。该算法通过嵌入含有单调递减性质的相对光滑H函数来逼近原来阶跃的比较函数S，并考虑了圆域内中心点和圆周邻点的灰度相似距离，以有区别地对待圆域内的灰度信息。另外，在该算法中增加了一个计数策略，以淘汰冗余的微小灰度变化。H-LBP计算简单，可以较为快速地完成边缘的提取；保留了对较大灰度变化的敏感性，增强了对灰度差异程度的描述能力，可以有区别地提取铸件边缘、灰度对比度较弱的小缺陷边缘和灰度不均的铸件号边缘，克服了噪声、伪影和图像模糊。H-LBP也可用于CT图像、背景复杂的图像和光照不均的图像。
本论文针对X射线图像特点和实际检测要求，对LBP作出了相应的改进。实际实验证实这些方法具有处理时间短和检测效果好两重优势，可有效地增强图像、提取边缘和识别缺陷，对工业X射线无损检测具有指导意义。
X-ray imaging techniques such as computed tomography (CT) and digitalradiography (DR), are widely been used on industrial inspetion and medical inprobing,as they are capable of showing the inner structure of tested object via images withoutdestructing object. In industrial applications, there are some time requests onnon-destructive detection by X-ray, especially on realtime detection, which ask for a fastdetection method. In DR image and CT volume data, the grayscale contrast ratios ofcastings and the distribution of defects are presented with local properties, which ask fora local information extraction method. However, traditional detecting methods are toughto keep the balance between processing time and final vision, due to the blurry and largesize of images.
In order to improve the efficiency of detection, local binary pattern (LBP), atexture discribtor, is exploted in X-ray images’ relative procession such as imageenhancement, edge extraction and feature extraction. LBP that has properties ascalculation simplicity, information locality, and monotic grayscale invariance, describesthe texture by calaulating the weighted sum of binary codes indicating the grayscaledifference in circular neighborhood. Acording to the traits of DR images and CT volumedata, and to profit from these advantages, LBP is improved respecting to aspects of theway to assign weights, the data to be handled, and the function to compare by. Suchimprovements has heightened the extraction capability of direction information andspatial information, improved the capture capability of edge information, andsuperinduced the description capability to the difference degree of grayscale.
Defects in castiings’ DR images are not very clear due to the low grayscale contrastratio. By utilizing the information locality, LBP can be used to enhance DR sub-imagewith defects, however, it cannot enhance DR sub-image with defects appearing indifferent directions. To overcome this shortcoming, the weighted sum calculation ofLBP is improved by using finite line integral transform (FLIT) to determine the maindirection of sub-image then to choose the right weight arrangement for the sumcalculation in LBP. The improved method is so-called FLIT-LBP. Involving thedirection property of FLIT, FLIT-LBP can enhance defects in different directions;Reserving the local advantages and grayscale invariance of LBP, FLIT-LBP can enhancedefects under different grayscale contrast ratios, especially defects in low contrast. Inaddition, after enhancement of FLIT-LBP, texture features of defects are emphasized so that relative texures can achieve a better recognition rate.
As CT volume data are conmmonly of large size, LBP is used to extract3D edgeby taking advantage of its calculation simplicity. Nevertheless, LBP is too sensitive tomicaro-changes in grayscale. In this case, we have studied a3D edge extraction methodcalled FL-fusion for CT volume data, by fusing spatial information of FLIT and LBP.Multiple direction information obtained by FLIT is fused to get binarizate volume data,so to satisfy the piecewise constant request of data in hand. Then multiple planeinformation obtained by LBP is fused to extract3D edges. FL-fusion method can obtainthin, continuous, occlusive and accurate3D edges without being affected by noises andartifacts, in a short time. A sifnifficant point is that FL-fusion also can maintain themicro feature of edges such as the tine of crack, which provides a base stone forparticular compoment analysis. This method can be used on kinds of CT volume data,whenever the inner structure simple or complicate.
Since DR images are blurry and large, LBP can be used to extract the edge due toits sensitivity to non-monotic grayscale changes. However, unable to describe thedegree of grayscale difference, LBP cannot distinguish randontant micro-changes anduseful changes of grayscale. To settle this problem, LBP is improved with a smoothcompare fuction. The so-called H-LBP embeds a comparative smooth and mononicdecreasing H funcrion into LBP to approaching original stepwise S function, andconsiders a similarity distance between surrounding point and centric point to describegrayscale relationships distinctively. Besides, a counting scheme is utilized to eliminateredundant micro-changes. With such modifications, H-LBP keeps the calculationsimplicity and maintains the sensitivity to larger grayscale changes while improves thedescription capability of grayscale difference degree. Thus, H-LBP is able to extractintegrated edges in blurry DR images, including small defect edges under low contrastratio and batch number edges under grayscale inhomogeneous conditions. Ourexperiments also show that H-LBP is applicable for CT images, complicate backgroundimages and illuminate inhomogeneous images.
To sum up, LBP is improved according to the traits of X-ray images and therequests of detection. These improved methods have advantages in both processing timeand detecting vision, feasible for image enhancement, edge extraction or defectclassification. Thus, our work plays a guiding role in indrustrial X-ray nondestructivedetection.
引文
[1]李博，宋文爱，林敏．X射线胶片图像特征提取和量化方法[J]．测试技术学报，2003，17（2）：118－121．
[2]孔凡琴，贾磊．基于加权减影的非等厚工件的X射线检测[J]．光学技术，2003，29（5）：590－594．
[3]张晓光，林家骏．X射线检测焊缝的图像处理与缺陷识别[J]．华东理工大学学报，2004，30（2）：199－202．
[4]孙岩，原培新．图像处理在X射线胶片缺陷识别中的应用[J]．机械工程与自动化，2006，（5）：61－63．
[5]蒋益女，徐从富．结合纹理分析和SVM的苹果梗蒂和缺陷识别方法[J]．计算机工程与应用，2008，44（1）：235－237．
[6]杨欧，郭宝平．胡涛．郭轩．灰度形态学在印刷品缺陷检测中的应用[J]．微计算机信息：测控自动化，2008，24（3－1）：226－227．
[7]宋强，徐科，徐金梧．基于结构谱的中厚板表面缺陷识别方法[J]．北京科技大学学报，2007，29（3）：342－345．
[8]熊凌，杨岚，陶诗锦，周俊．基于图像处理的钢帘线表面缺陷特征分析[J]．湖北工业大学学报，2008，23（2）：26－29．
[9]那彦，焦李成．基于多分辨分析理论的图像融合方法[M]．西安：西安电子科技大学出版社，2007．
[10]焦李成，张向荣，侯彪，王爽，刘芳．智能SAR图像处理与解译[M]．北京：科技出版社，2008．
[11] X．P．He，P．Han，X．G．Lu，R．B．Wu．A new enhancement technique of X-ray carry-onluggage images based on DWT and fuzzy theory [C]．International Conference on ComputerScience and Information Technology，2008：855－858．
[12] Q．Bing，J．Lu，Z．Jing．A novel image enhancement algorithm based on information fusion[C]．International Conference on Computer Science and Software Engineering，2008：577－580．
[13] Q．G．Miao and B．S．Wang． A novel image fusion algorithm using FRIT and PCA [C]．10thInternational Conference on Information Fusion，2007：1－5．
[14] J．Li，Q．Pan，T．Yang，Y．M．Cheng．Color based grayscale-fused enhancement algorithmfor video surveillance [C]．Proceedings of the Third International Conference on Image andGraphics，2004：47－50．
[15] L．Tao，H．Ngo，M．Zhang，A．Livingston，V．Asari．A multi-sensor image fusion andenhancement system for assisting drivers in poor lighting conditions [C]．Proceedings of the34th Applied Imagery and Pattern Recognition Workshop，2005：106－111．
[16] G．Z．Li，W．S．Luo，P．Li，H．B．Lv．Fusion enhancement of color image based onglobal histogram equalization [C]．International Conference on Computer Science andSoftware Engineering，2008：205－208．
[17] X．Q．Zhang．Infrared and color visible image sequence fusion based on statistical model andimage enhancement [C]．International Conference on Advanced Computer Theory andEngineering，2008：934－937．
[18] F．E．Ali，I．M．El-Dokany，A．A．Saad，F．E．Abd El-Samie．Fusion of MR and CTimages using the curvelet transformation [C]．25th National Radio Science Conference，2008，3：1－8．
[19] T．Chen，J．P．Zhang，Y．Zhang．Remote sensing image fusion based on ridgelet transform[C]．International Geoscience and Remote Sensing Symposium，2005：1150－1153．
[20] F．Liu，J．Y．Liu，Y．Gao．Image fusion based on wadgelet and wavelet [C]．Proceedingof2007International Symposium on Intelligent Signal Processing and CommunicationSystems，2007：682－685．
[21] Y．Chai，Y．He，C．L．Ying．CT and MRI image fusion based on contourlet using a novelrule [C]．2ndInternational Conference on Bioinformatics and Biomedical Engineering，2008：2064－2067．
[22] B．Yang，S．T．Li，F．M．Sun．Image fusion using nonsubsampled contourlet transform[C]．Fourth International Conference on Image and Graphics，2007：719－724．
[23] L．Yang，B．L．Guo，W．Ni．Multifocus image fusion algorithm based on contourletdecomposition and region statistics [C]．Fourth International Conference on Image andGraphics，2007：707－711．
[24]焦李成，侯彪，王爽，刘芳．图像多尺度几何分析利用与应用——后小波分析理论和应用[M]．西安电子科技大学出版社，2008．
[25]马睿．工业CT高精度图像测量算法研究[D]．重庆：重庆大学，2008．
[26]郭海燕．图像分割算法研究及在工业CT中的应用[D]．重庆：重庆大学，2009．
[27]向才兵．基于CV模型的工业CT图像裂纹检测算法研究[D]．重庆：重庆大学，2010．
[28]马睿，曾理，卢艳平．改进的基于Facet模型的亚像素边缘检测[J]．应用基础与工程科学学报：2009，17（2）：298－301．
[29]曾理，悦秀娟，侯立华．基于改进Snake模型的工业CT图像内窥显示及测量[J]．计算机应用：2007，27（1）：177－179．
[30]刘玲慧．工业CT/DR图像缺陷检测算法研究[D]．重庆：重庆大学，2010．
[31]安贝贝．工业CT图像缺陷检测的脊波算法研究[D]．重庆：重庆大学，2007．
[32]曾理，安贝贝，马睿．脊波在工业CT图像裂纹边缘检测中的应用[J]．仪器仪表学报，2007，28（6）：981－986．
[33]悦秀娟．基于改进脊波变换的工业CT图像裂纹检测[D]．重庆：重庆大学，2009．
[34]李宗剑．基于多尺度几何分析的工业CT图像和体数据特征提取的研究[D]．重庆：重庆大学，2010．
[35]龙奕，刘明坤，尹忠科，王建英．基于Bandelet变换域的图像自适应增强算法[J]．计算机应用，2008，28（5）：1221－1224．
[36]侯彪，徐婧，刘凤，焦李成．基于第二代Bandelet域隐马尔科夫树模型的图像分割[J]．自动化学报，2009，35（5）：498－504．
[37] S．Yang，M．Wang，L．Jiao．Quantum-inspired immune clone algorithm and multiscaleBandelet based image representation [J]．Pattern Recognition Lett．，2010，31（13）：1894－1902．
[38] S．Yang，Y．Liu，M．Wang，L．Jiao．Low bit rate SAR image coding based on adaptivemultiscale Bandelets and cooperative decision [J]．Signal Processing，2009，89：1910－1920．
[39]张秀英，冯学智，江洪．面向对象分类的特征空间优化[J]．遥感学报，2009，13（4）：664－668．
[40] R．R．Da Silva，L．P．Caloba，M．H．S．Siqueir，J．M．A．Rebello．Pattern recognitionof weld defects detected by radiographic test [J]．NDT&E International2004，37：461－470．
[41] X．G．Zhang，J．J．Xu，Y．Li．The research of defect recognition for radiographic weldimage based on fuzyy neural network [C]．Proceeding of the5th World Congress onIntelligent Control and Automation，2004，6：2661－2665．
[42] X．G．Zhang，J．J．Xu，G．Y．Ge．Defects recognition on X-ray images for weld inspectionusing SVM [C]．Proceeding of the Third International Conference on Machine Learning andCybernetics，2004：3721－3725．
[43] T．W．Liao．Classification of weld flaws with imbalanced class data [J]．Expert Systems withApplications，2008，35：1041－1052.
[44] H．I．Shafeek，E．S．Gadelmawla，A．A．Abdel-Shafy，I．M．Elewa．Automatic inspectionof gas pipeline welding defects using an expert vision system [J]．NDT&E International，2004，37：301－307．
[45]沈宽．铸件数字辐射图像的缺陷自动识别研究[D]．重庆：重庆大学，2010．
[46] T．Celik and T．Tjahjadi．Multiscale texture classification using dual-tree complex wavelettransform [J]．Pattern Recognition Letters，2009，30：331－339．
[47] S．Kumar，S．H．Ong，S．Ranganath，F．T．Chew．Invariant texture classification forbiomedical cell specimens via non-linear polar map filtering [J]．Computer Vision and ImageUnderstanding，2010，114：44－53．
[48] T．Ojala，M．Pietikainen，D．Harwood．A comparative study of texture measures withclassification based on feature distributions [J]．Pattern Recognition，1996，29（1）：51－59．
[49] M．Pietikainen，T．Ojala，Z．Xu．Rotation-invariant texture classification using featuredistributions [J]．Pattern Recognition，2000，33：43－52．
[50] T．Ojala，M．Pietikainen，T．Maenpaa．Multiresolutioon gray-scale and rotation invarianttxture classification with local binary patterns [J]．IEEE transactions on pattern analysis andmachine intelligence，2002，24（7）：971－987．
[51] T．Ahonen，A．hadid，M．Pietikainen．Face recognition with local binary patterns[C]．Proceedings of the European Conference on Computer Vision，2004，LNCS3021：469－481．
[52] L．H．He，C．R．Zou，L．Zhao，D．Hu．An enhanced LBP feature based on facial expressionrecognition [C]．Proceeding of the2005IEEE Engineering in Medicine and Biology27thAnnual Conference，2005：3300－3303．
[53] N．Tan，L．Huang，C．H．Liu．Face recognition based on a single image and local binarypattern [C]．Intelligent Information Technology Application，2008，2：366－370．
[54] N．Tan，L．Huang，C．H．Liu．Face recognition based on LBP and orthogonal Rank-Onetensor projections[C]．International Conference on Pattern Recognition，2008：1－4．
[55] Z．L．Ying and L．B．Cai．Support vector discriminant analysis on local binary patterns forfacial expression recognition[C]．2ndInternational Conference on Image and SignalProcessing，2009：1－4．
[56]王玮，黄非非，李见为，冯海亮．使用多尺度LBP特征描述与识别人脸[J]．光学精密工程，2008，16（4）：696－705．
[57]高涛，何明一，戴玉超，白磷．多级LBP直方图序列特征的人脸识别[J]．中国图象图形学报，2009，14（2）：202－207．
[58]刘中华，史恒亮，张兰萍，金忠．基于多尺度局部二值模式的人脸识别[J]．计算机科学，2009，36（11）：293－299．
[59] P．T．Pham-Ngoc and K．H．Jo．Color-based face detection using combination of modifiedlocal binary patterns and embedded hidden Markov models [C]．SICE-ICASE InternationalJoint Conference2006：5598－5603．
[60] S．Q．Zhao and Y．S．Gao．Establishing Point Correspondence Using Multidirectional BinaryPattern for Face Recognition [C]．19thInternational Conference on Pattern Recognition，2008：1－4．
[61] L．Setia，A．Teynor，A．Halawani，H．Burkhardt．Grayscale medical image annotationusing local relational features [J]．Pattern Recognition Letters，2008，29：2039－2045．
[62] Y．Wang，X．Y．Wei，S．Xiao．LBP texture analysisi based on the local adaptive Niblackalgorithm [C]．2008Congress on Image and Signal Processing，2008：777－780．
[63] X．Y．Tan and B．Triggs．Enhanced local texture feature sets for face recognition underdifficult lighting conditions[J]，IEEE Trans. Image Process．，2010，19：1635－1650．
[64]赵建民，朱信忠，江小辉．基于改进型LBP特征的人脸识别方法研究[J]．计算机科学，2009，36（8）：276－280．
[65] S．H．Peng，D．H．Kim，S．L．Lee，M．K.．Lim．Texture feature extraction based onuniformity estimation method for local brightness and structure in chest CT images[J]．Comput．Biol．Med．，2010，40：931－942．
[66]胡园园，王让定．基于局部纹理不变性的运动阴影去除算法[J]．计算机应用，2008，28（12）：3141－3143．
[67] L．Nanni，A．Lumini，S．Brahnam．Local binary patterns variants as texture descriptors formedical image analysis [J]，Artif．Intell．Med．，2010，49：117－125．
[68] H．Zhou，R．S．Wang，C．Wang．A novel extended local-binary-pattern operator for textureanalysis [J]．Information Sciences，2008，178：4314－4325．
[69]徐剑，丁晓青，王生进，吴佑寿．一种融合局部纹理和颜色信息的背景减除方法[J]．自动化学报，2009，35（9）：1145－1150．
[70] W．H．Ma，L．Huang，C．P．Liu．Advanced local binary pattern descriptors for crowdestimation [C]．IEEE Pacific-Asia Workshop on Computational Intelligence and IndustrialApplication，2008：958－962．
[71]齐永锋，张家树，火元连．一种基于肤色和改进的LBP的人脸检测方法[J]．光电子．激光，2009，20（6）：816－821．
[72]付晓峰，韦巍．基于多尺度中心化二值模式的人脸表情识别[J]．控制理论与应用，2009，26（6）：629－633．
[73] Y．C．Fang and Z．Wang．Improving LBP features for gender calssification [C]．Proceedingsof the2008International Conference on Wavelet Analysis and Pattern Recognition，HongKong，2008：373－377．
[74]刘民霞，侯迎坤，郭小春，张燕，杨德运．新的纹理图像特征提取方法[J]．计算机应用，2009，29（12）：3434－3436．
[75] A．Hadid and M．Pietikainen．Combining appearance and motion for face and genderrecognition from videos [J]．Pattern Recognition，2009，42：2818－2827．
[76] G．Zhao and M．Pietikainen．Dynamic texture recognition using local binary patterns with anapplication to facial expressions [J]．IEEE Transactions on Pattern Analysis and MachineIntelligence，2007，29：915－928．
[77] D．G．Lowe．Object recognition from local scale-invariant features [C]．Proceedings of IEEEInternational Conference of Computer Vision，1999，2：1150－1157．
[78] C．Belcher and Y．Z．Du．Region-based SIFT approach to iris recognition [J]．Optics andLasers in Engineering，2009，47：139－147．
[79] K．Mikolajczyk and C．Schmid．A performance evaluation of local descriptors [J]．IEEETransections on Pattern Analysis and Machie Intteligence，2005，27（10）：1615－1630．
[80] J．Li and N．M．Allinson．A comprehensive review of current local features for computervision [J]．Neurocomputing，2008，71：1771－1787．
[81]李建福．红外图像中人体目标检测、跟踪及行为识别研究[D]．重庆：重庆大学，2010．
[82] H．Ⅱ．Kim，S．H．Lee，N．IK Cho，Member，IEEE．Automatic defect classificationusing frequency and spatial features in a Boosting scheme [J]．IEEE Signal ProcessingLetters，2009，16（5）：374－377．
[83] S．C．Kim and T．J．Kang．Texture classification and segmentation using incomplete treestructured wavelet packet frame and Gaussian mixture model [C]．International Workshop onImaging Systems and Techniques，2005：46－51．
[84] L．Tomczak and V．Mosorov．Singular value decomposition for texture defect detection invisual inspection systems [C]．Proceedings of the2ndInternational Conference on Memstech，2006：131－133．
[85] Y．Wang，Y．Sun，P．Lv，H．Wang．Detection of line weld defects based on multiple thresholdsand support vector machine [J]．NDT&E International，2008，41：517－524．
[86] H．X．Sun，Y．H．Zhang，F．L．Luo．Texture defect detection of wire rope surface withsupport vector data description [C]．Chinese Conference on Pattern Recognition，2009：1－5．
[87] A．A．Carvalho，J．M．A．Rebello，L．V．S．Sagrilo，C．S．Camerini，I．V．J．Miranda．MFLsignals and artificial neural networks applied to detection and classification of pipe welddefects [J]．NDT&E International，2006，39：661－667．
[88] J．Mirapeix，P．B．Garcia-Allende，A．Cobo，O．M．Conde，J．M．Lopez-Higuera．Real-timearc-welding defect detection and classification with principal component analysis and articleneural networks [J]．NDT&E International，2007，40：315－323．
[89] A．A．Carvalho，J．M．A．Rebello，M．P．V．Souza，L．V．S．Sagrilo，S．D．Soares．Reliabilityof non-destructive test techniques in the inspection of pipelines used in the oilindustry[J]．International Journal of Pressure Vessels and Piping，2008，85：745－751．
[90] P．B．Garcia-Allende，J．Mirapeix，O．M．Conde，A．Cobo，J．M．Lopea-Higuera．Spectralprocessing technique based on feature selection and artificial neural networks for arc-weldingquality monitoring [J]．NDT&E International，2009，42：56－63．
[91] R．Vilar，J．Zapata，R．Ruiz．An automatic system of classification of weld defects inradiographic images [J]．NDT&E International，2009，42：467－476．
[92]张延林,谢永华．木材缺陷识别方法的研究[J]．森林工程，2008，24（4）：33－35．
[93] Y．Zhang，W．S．Cheng，J．Zhao．Classification of surface Detects of Strips Based onInvariable Moment Functions [J]．Opto-Electronic Engineering，2008，35（7）：90－94．
[94] D．Gao，Y．X．Liu，X．G．Zhang，Y．X．Liu．Binary-tree multi-classifier for weldingdefects and its application based on SVM [C]．Proceeding of the8th World Congress onIntelligent Control and Automation，2006：8509-8513．
[95]孙林，杨世元，吴德会．X射线底片焊缝缺陷的支持向量机识别方法[J]．应用科学学报，2008，26（4）：418－424．
[96] H．G．Bu，J．Wang，X．B．Huang．Fabric defect detection based on multiple fractal featuresand support vector data description [J]．Engineering Applications of Artificial Intelligence，2009，22：224－235．
[97] T．S．Li．Applying wavelets transforms，rough set theory and support vector machine forcopper clad laminate defects classification [J]．Exports Systems with Applications，2009，36：5822－5829．
[98]贺秋伟，王龙山，于忠党，李国发，高立国．基于图像处理和支持向量机的微型齿轮缺陷检测[J]．吉林大学学报，2008，38（5）：565－569．
[99] T．W．Liao，A．K．Celmins，R．J．Hammell Ⅱ．A fuzzy c-means variant for the generationof fuzzy term sets [J]．Fuzzy Sets and System，2003，135：241－257．
[100] T．W．Liao．Classification of welding flaw types with fuzzy expert systems [J]．ExpertSystems with Application，2003，25：101－111．
[101]李永生，戚大伟．基于FCM的木材X射线图像缺陷分割[J]．森林工程，2008，24（3）：51－52．
[102] F．Herold，K．Bavendiek，R．Grigat．A Third Generation Automatic Defect Detection System[R]．
[103] L．Qin，Q．F．Zheng，et al．Unsupervised texture classification：Automatically discoverand classify texture patterns [J]．Image and Vision Computing，2008，26：647－656．
[104] Y．L．Qiao，Z．M．Lu，J．S．Pan，S．H．Sun．Fast k-nearest neighbor search algorithmbased on pyramid structure of wavelet transform and its application to texture classification[J]．Digital Signal Processing：A Review Journal，2010，20（3）：837－845．
[105] D．Le and S．Satoh．Ent-Boost：Boosting using entropy measures for robust object detection[J]．Pattern Recognition Letters，2007，28：1083－1090．
[106]林开颜，吴军辉．基于计算机视觉的水果分级技术研究进展[J]．信息化纵横，2009，（10）：1－9．
[107]陈小娜，章程辉．绿橙表面缺陷的计算机视觉分级技术[J]．农机化研究，2009，（10）：126－129．
[108]龚毅光，白俊杰，王宁生．汽车产品审核中智能评价缺陷技术的研究与应用[J]．机械科学与技术，2008，27（12）：1552－1558．
[109]孙朝明，汤光平，李强，王增勇．数字化射线检测图像的缺陷评级方法[C]．全国射线检测技术年会论文集，2011：75－81．
[110]徐兴林，贾国华．双臂双投影法透照椭圆一次成像在小径管焊口缺陷评级中的使用[J]．热力发电，2005，（6）：50－54．
[111]庄天戈．CT理论与算法[M]．上海：上海交通大学出版社，1992．
[112] G．L．Zeng．Medical Image Reconstruction [M]．Springer，2009．
[113]叶海霞．工业CT窄角扇束卷积反投影并行图像重建研究[D]．重庆：重庆大学，2003．
[114]叶云长．计算机层析成像检测[M]．北京：机械工业出版社，2006．
[115] http://www.volren.org/[EB/OL]．
[116] M．Q．Yang，Y．H．Peng，Y．X．Liu．The algorithm and application of finite line integraltransform[C]．IEEE International Symposium on Microwave，Antenna，Propagation and EMCTechnologies Proceedings，2005：411－414．
[117] K．I．Laws．Texture image segmentation [R]．Image Processing Institute，Univ. of SouthernCalifornia，1980．
[118] S．Q．Zhao，Y．S．Gao，B．C．Zhang．Sobel-LBP [C]．In：IEEE Internat．Conf．onImage Proces．（ICIP），2008：2144－2147．
[119] Q．Wang，J．M．Liang，et al．．Yang．Spatial texture based automatic classification of daysideaurora in all-sky image [J]．J．Atmos．Sol．-Terr．Phys．，2010，72：498－508．
[120] M．M．Azab，H．A．Shedeed，A．S．Hussein．A new technique for background modelingand subtraction for motion detection in real-time videos [C]．IEEE Internat．Conf．on ImageProces．（ICIP），2010：3453－3456．
[121] H．Sun，C．Wang，B．L．Wang，N．EL-Sheimy．Pyramid binary pattern features for real-timepedestrian detection from infrared videos [J]．Neurocomputing，2011，74：797－804．
[122] K．B．Wang，B．Z．Yu，W．Xi．Radar image segmentation using active contour [C]．20071st Asian and Pacific Conference on Synthetic Aperture Radar Pages，2007：446－450．
[123] Y．Li，S．Luo，Q．Zou．Learning to detect boundaries in natural image using texture cuesand EM [C]．2008Fourth International Conference on Natural Computation（ICNC），2008：167－171．
[124] M．A．Savelonas，D．K．Iakovidis，D．Maroulis．LBP－guided active contours [J]．PatternRecognit．Lett．，2008，29：1404－1415．
[125]曾理，蒲云，马睿．基于工业CT的铁路货车铸件缺陷自动检测[J]．中国铁道科学，2009，30（4）：76－80．
[126] H．L．Jin，Q．S．Liu，et al．．Face detection using improved LBP under Bayesian framework
[C]．Proceedings. Third International Conference on Image and Graphics，2004：306－309｜xvii+588．
[127] Y．Wang，X．Y．Wei，S．Xiao．LBP texture analysis based on the local adaptive Niblackalgorithm [C]．CISP2008：Proceedings of First International Congress on Image and SignalProcessing，2008，2：777－780．
[128] G．D．Tian and A．D．Men．An improved texture-based method for background subtractionusing local binary patterns [C]．Proceedings of The20092ndInternational Congress on Imageand Signal Processing，2009，1－9：1984－1987．
[129] http://www.3dmed.net/[EB/OL]．
[130] L．Wang，L．He，A．Mishra，C．M．Li．Active contours driven by local Gaussian distributionfitting energy [J]．Signal Process，2009，89：2435－2447．
[2]孔凡琴，贾磊．基于加权减影的非等厚工件的X射线检测[J]．光学技术，2003，29（5）：590－594．
[3]张晓光，林家骏．X射线检测焊缝的图像处理与缺陷识别[J]．华东理工大学学报，2004，30（2）：199－202．
[4]孙岩，原培新．图像处理在X射线胶片缺陷识别中的应用[J]．机械工程与自动化，2006，（5）：61－63．
[5]蒋益女，徐从富．结合纹理分析和SVM的苹果梗蒂和缺陷识别方法[J]．计算机工程与应用，2008，44（1）：235－237．
[6]杨欧，郭宝平．胡涛．郭轩．灰度形态学在印刷品缺陷检测中的应用[J]．微计算机信息：测控自动化，2008，24（3－1）：226－227．
[7]宋强，徐科，徐金梧．基于结构谱的中厚板表面缺陷识别方法[J]．北京科技大学学报，2007，29（3）：342－345．
[8]熊凌，杨岚，陶诗锦，周俊．基于图像处理的钢帘线表面缺陷特征分析[J]．湖北工业大学学报，2008，23（2）：26－29．
[9]那彦，焦李成．基于多分辨分析理论的图像融合方法[M]．西安：西安电子科技大学出版社，2007．
[10]焦李成，张向荣，侯彪，王爽，刘芳．智能SAR图像处理与解译[M]．北京：科技出版社，2008．
[11] X．P．He，P．Han，X．G．Lu，R．B．Wu．A new enhancement technique of X-ray carry-onluggage images based on DWT and fuzzy theory [C]．International Conference on ComputerScience and Information Technology，2008：855－858．
[12] Q．Bing，J．Lu，Z．Jing．A novel image enhancement algorithm based on information fusion[C]．International Conference on Computer Science and Software Engineering，2008：577－580．
[13] Q．G．Miao and B．S．Wang． A novel image fusion algorithm using FRIT and PCA [C]．10thInternational Conference on Information Fusion，2007：1－5．
[14] J．Li，Q．Pan，T．Yang，Y．M．Cheng．Color based grayscale-fused enhancement algorithmfor video surveillance [C]．Proceedings of the Third International Conference on Image andGraphics，2004：47－50．
[15] L．Tao，H．Ngo，M．Zhang，A．Livingston，V．Asari．A multi-sensor image fusion andenhancement system for assisting drivers in poor lighting conditions [C]．Proceedings of the34th Applied Imagery and Pattern Recognition Workshop，2005：106－111．
[16] G．Z．Li，W．S．Luo，P．Li，H．B．Lv．Fusion enhancement of color image based onglobal histogram equalization [C]．International Conference on Computer Science andSoftware Engineering，2008：205－208．
[17] X．Q．Zhang．Infrared and color visible image sequence fusion based on statistical model andimage enhancement [C]．International Conference on Advanced Computer Theory andEngineering，2008：934－937．
[18] F．E．Ali，I．M．El-Dokany，A．A．Saad，F．E．Abd El-Samie．Fusion of MR and CTimages using the curvelet transformation [C]．25th National Radio Science Conference，2008，3：1－8．
[19] T．Chen，J．P．Zhang，Y．Zhang．Remote sensing image fusion based on ridgelet transform[C]．International Geoscience and Remote Sensing Symposium，2005：1150－1153．
[20] F．Liu，J．Y．Liu，Y．Gao．Image fusion based on wadgelet and wavelet [C]．Proceedingof2007International Symposium on Intelligent Signal Processing and CommunicationSystems，2007：682－685．
[21] Y．Chai，Y．He，C．L．Ying．CT and MRI image fusion based on contourlet using a novelrule [C]．2ndInternational Conference on Bioinformatics and Biomedical Engineering，2008：2064－2067．
[22] B．Yang，S．T．Li，F．M．Sun．Image fusion using nonsubsampled contourlet transform[C]．Fourth International Conference on Image and Graphics，2007：719－724．
[23] L．Yang，B．L．Guo，W．Ni．Multifocus image fusion algorithm based on contourletdecomposition and region statistics [C]．Fourth International Conference on Image andGraphics，2007：707－711．
[24]焦李成，侯彪，王爽，刘芳．图像多尺度几何分析利用与应用——后小波分析理论和应用[M]．西安电子科技大学出版社，2008．
[25]马睿．工业CT高精度图像测量算法研究[D]．重庆：重庆大学，2008．
[26]郭海燕．图像分割算法研究及在工业CT中的应用[D]．重庆：重庆大学，2009．
[27]向才兵．基于CV模型的工业CT图像裂纹检测算法研究[D]．重庆：重庆大学，2010．
[28]马睿，曾理，卢艳平．改进的基于Facet模型的亚像素边缘检测[J]．应用基础与工程科学学报：2009，17（2）：298－301．
[29]曾理，悦秀娟，侯立华．基于改进Snake模型的工业CT图像内窥显示及测量[J]．计算机应用：2007，27（1）：177－179．
[30]刘玲慧．工业CT/DR图像缺陷检测算法研究[D]．重庆：重庆大学，2010．
[31]安贝贝．工业CT图像缺陷检测的脊波算法研究[D]．重庆：重庆大学，2007．
[32]曾理，安贝贝，马睿．脊波在工业CT图像裂纹边缘检测中的应用[J]．仪器仪表学报，2007，28（6）：981－986．
[33]悦秀娟．基于改进脊波变换的工业CT图像裂纹检测[D]．重庆：重庆大学，2009．
[34]李宗剑．基于多尺度几何分析的工业CT图像和体数据特征提取的研究[D]．重庆：重庆大学，2010．
[35]龙奕，刘明坤，尹忠科，王建英．基于Bandelet变换域的图像自适应增强算法[J]．计算机应用，2008，28（5）：1221－1224．
[36]侯彪，徐婧，刘凤，焦李成．基于第二代Bandelet域隐马尔科夫树模型的图像分割[J]．自动化学报，2009，35（5）：498－504．
[37] S．Yang，M．Wang，L．Jiao．Quantum-inspired immune clone algorithm and multiscaleBandelet based image representation [J]．Pattern Recognition Lett．，2010，31（13）：1894－1902．
[38] S．Yang，Y．Liu，M．Wang，L．Jiao．Low bit rate SAR image coding based on adaptivemultiscale Bandelets and cooperative decision [J]．Signal Processing，2009，89：1910－1920．
[39]张秀英，冯学智，江洪．面向对象分类的特征空间优化[J]．遥感学报，2009，13（4）：664－668．
[40] R．R．Da Silva，L．P．Caloba，M．H．S．Siqueir，J．M．A．Rebello．Pattern recognitionof weld defects detected by radiographic test [J]．NDT&E International2004，37：461－470．
[41] X．G．Zhang，J．J．Xu，Y．Li．The research of defect recognition for radiographic weldimage based on fuzyy neural network [C]．Proceeding of the5th World Congress onIntelligent Control and Automation，2004，6：2661－2665．
[42] X．G．Zhang，J．J．Xu，G．Y．Ge．Defects recognition on X-ray images for weld inspectionusing SVM [C]．Proceeding of the Third International Conference on Machine Learning andCybernetics，2004：3721－3725．
[43] T．W．Liao．Classification of weld flaws with imbalanced class data [J]．Expert Systems withApplications，2008，35：1041－1052.
[44] H．I．Shafeek，E．S．Gadelmawla，A．A．Abdel-Shafy，I．M．Elewa．Automatic inspectionof gas pipeline welding defects using an expert vision system [J]．NDT&E International，2004，37：301－307．
[45]沈宽．铸件数字辐射图像的缺陷自动识别研究[D]．重庆：重庆大学，2010．
[46] T．Celik and T．Tjahjadi．Multiscale texture classification using dual-tree complex wavelettransform [J]．Pattern Recognition Letters，2009，30：331－339．
[47] S．Kumar，S．H．Ong，S．Ranganath，F．T．Chew．Invariant texture classification forbiomedical cell specimens via non-linear polar map filtering [J]．Computer Vision and ImageUnderstanding，2010，114：44－53．
[48] T．Ojala，M．Pietikainen，D．Harwood．A comparative study of texture measures withclassification based on feature distributions [J]．Pattern Recognition，1996，29（1）：51－59．
[49] M．Pietikainen，T．Ojala，Z．Xu．Rotation-invariant texture classification using featuredistributions [J]．Pattern Recognition，2000，33：43－52．
[50] T．Ojala，M．Pietikainen，T．Maenpaa．Multiresolutioon gray-scale and rotation invarianttxture classification with local binary patterns [J]．IEEE transactions on pattern analysis andmachine intelligence，2002，24（7）：971－987．
[51] T．Ahonen，A．hadid，M．Pietikainen．Face recognition with local binary patterns[C]．Proceedings of the European Conference on Computer Vision，2004，LNCS3021：469－481．
[52] L．H．He，C．R．Zou，L．Zhao，D．Hu．An enhanced LBP feature based on facial expressionrecognition [C]．Proceeding of the2005IEEE Engineering in Medicine and Biology27thAnnual Conference，2005：3300－3303．
[53] N．Tan，L．Huang，C．H．Liu．Face recognition based on a single image and local binarypattern [C]．Intelligent Information Technology Application，2008，2：366－370．
[54] N．Tan，L．Huang，C．H．Liu．Face recognition based on LBP and orthogonal Rank-Onetensor projections[C]．International Conference on Pattern Recognition，2008：1－4．
[55] Z．L．Ying and L．B．Cai．Support vector discriminant analysis on local binary patterns forfacial expression recognition[C]．2ndInternational Conference on Image and SignalProcessing，2009：1－4．
[56]王玮，黄非非，李见为，冯海亮．使用多尺度LBP特征描述与识别人脸[J]．光学精密工程，2008，16（4）：696－705．
[57]高涛，何明一，戴玉超，白磷．多级LBP直方图序列特征的人脸识别[J]．中国图象图形学报，2009，14（2）：202－207．
[58]刘中华，史恒亮，张兰萍，金忠．基于多尺度局部二值模式的人脸识别[J]．计算机科学，2009，36（11）：293－299．
[59] P．T．Pham-Ngoc and K．H．Jo．Color-based face detection using combination of modifiedlocal binary patterns and embedded hidden Markov models [C]．SICE-ICASE InternationalJoint Conference2006：5598－5603．
[60] S．Q．Zhao and Y．S．Gao．Establishing Point Correspondence Using Multidirectional BinaryPattern for Face Recognition [C]．19thInternational Conference on Pattern Recognition，2008：1－4．
[61] L．Setia，A．Teynor，A．Halawani，H．Burkhardt．Grayscale medical image annotationusing local relational features [J]．Pattern Recognition Letters，2008，29：2039－2045．
[62] Y．Wang，X．Y．Wei，S．Xiao．LBP texture analysisi based on the local adaptive Niblackalgorithm [C]．2008Congress on Image and Signal Processing，2008：777－780．
[63] X．Y．Tan and B．Triggs．Enhanced local texture feature sets for face recognition underdifficult lighting conditions[J]，IEEE Trans. Image Process．，2010，19：1635－1650．
[64]赵建民，朱信忠，江小辉．基于改进型LBP特征的人脸识别方法研究[J]．计算机科学，2009，36（8）：276－280．
[65] S．H．Peng，D．H．Kim，S．L．Lee，M．K.．Lim．Texture feature extraction based onuniformity estimation method for local brightness and structure in chest CT images[J]．Comput．Biol．Med．，2010，40：931－942．
[66]胡园园，王让定．基于局部纹理不变性的运动阴影去除算法[J]．计算机应用，2008，28（12）：3141－3143．
[67] L．Nanni，A．Lumini，S．Brahnam．Local binary patterns variants as texture descriptors formedical image analysis [J]，Artif．Intell．Med．，2010，49：117－125．
[68] H．Zhou，R．S．Wang，C．Wang．A novel extended local-binary-pattern operator for textureanalysis [J]．Information Sciences，2008，178：4314－4325．
[69]徐剑，丁晓青，王生进，吴佑寿．一种融合局部纹理和颜色信息的背景减除方法[J]．自动化学报，2009，35（9）：1145－1150．
[70] W．H．Ma，L．Huang，C．P．Liu．Advanced local binary pattern descriptors for crowdestimation [C]．IEEE Pacific-Asia Workshop on Computational Intelligence and IndustrialApplication，2008：958－962．
[71]齐永锋，张家树，火元连．一种基于肤色和改进的LBP的人脸检测方法[J]．光电子．激光，2009，20（6）：816－821．
[72]付晓峰，韦巍．基于多尺度中心化二值模式的人脸表情识别[J]．控制理论与应用，2009，26（6）：629－633．
[73] Y．C．Fang and Z．Wang．Improving LBP features for gender calssification [C]．Proceedingsof the2008International Conference on Wavelet Analysis and Pattern Recognition，HongKong，2008：373－377．
[74]刘民霞，侯迎坤，郭小春，张燕，杨德运．新的纹理图像特征提取方法[J]．计算机应用，2009，29（12）：3434－3436．
[75] A．Hadid and M．Pietikainen．Combining appearance and motion for face and genderrecognition from videos [J]．Pattern Recognition，2009，42：2818－2827．
[76] G．Zhao and M．Pietikainen．Dynamic texture recognition using local binary patterns with anapplication to facial expressions [J]．IEEE Transactions on Pattern Analysis and MachineIntelligence，2007，29：915－928．
[77] D．G．Lowe．Object recognition from local scale-invariant features [C]．Proceedings of IEEEInternational Conference of Computer Vision，1999，2：1150－1157．
[78] C．Belcher and Y．Z．Du．Region-based SIFT approach to iris recognition [J]．Optics andLasers in Engineering，2009，47：139－147．
[79] K．Mikolajczyk and C．Schmid．A performance evaluation of local descriptors [J]．IEEETransections on Pattern Analysis and Machie Intteligence，2005，27（10）：1615－1630．
[80] J．Li and N．M．Allinson．A comprehensive review of current local features for computervision [J]．Neurocomputing，2008，71：1771－1787．
[81]李建福．红外图像中人体目标检测、跟踪及行为识别研究[D]．重庆：重庆大学，2010．
[82] H．Ⅱ．Kim，S．H．Lee，N．IK Cho，Member，IEEE．Automatic defect classificationusing frequency and spatial features in a Boosting scheme [J]．IEEE Signal ProcessingLetters，2009，16（5）：374－377．
[83] S．C．Kim and T．J．Kang．Texture classification and segmentation using incomplete treestructured wavelet packet frame and Gaussian mixture model [C]．International Workshop onImaging Systems and Techniques，2005：46－51．
[84] L．Tomczak and V．Mosorov．Singular value decomposition for texture defect detection invisual inspection systems [C]．Proceedings of the2ndInternational Conference on Memstech，2006：131－133．
[85] Y．Wang，Y．Sun，P．Lv，H．Wang．Detection of line weld defects based on multiple thresholdsand support vector machine [J]．NDT&E International，2008，41：517－524．
[86] H．X．Sun，Y．H．Zhang，F．L．Luo．Texture defect detection of wire rope surface withsupport vector data description [C]．Chinese Conference on Pattern Recognition，2009：1－5．
[87] A．A．Carvalho，J．M．A．Rebello，L．V．S．Sagrilo，C．S．Camerini，I．V．J．Miranda．MFLsignals and artificial neural networks applied to detection and classification of pipe welddefects [J]．NDT&E International，2006，39：661－667．
[88] J．Mirapeix，P．B．Garcia-Allende，A．Cobo，O．M．Conde，J．M．Lopez-Higuera．Real-timearc-welding defect detection and classification with principal component analysis and articleneural networks [J]．NDT&E International，2007，40：315－323．
[89] A．A．Carvalho，J．M．A．Rebello，M．P．V．Souza，L．V．S．Sagrilo，S．D．Soares．Reliabilityof non-destructive test techniques in the inspection of pipelines used in the oilindustry[J]．International Journal of Pressure Vessels and Piping，2008，85：745－751．
[90] P．B．Garcia-Allende，J．Mirapeix，O．M．Conde，A．Cobo，J．M．Lopea-Higuera．Spectralprocessing technique based on feature selection and artificial neural networks for arc-weldingquality monitoring [J]．NDT&E International，2009，42：56－63．
[91] R．Vilar，J．Zapata，R．Ruiz．An automatic system of classification of weld defects inradiographic images [J]．NDT&E International，2009，42：467－476．
[92]张延林,谢永华．木材缺陷识别方法的研究[J]．森林工程，2008，24（4）：33－35．
[93] Y．Zhang，W．S．Cheng，J．Zhao．Classification of surface Detects of Strips Based onInvariable Moment Functions [J]．Opto-Electronic Engineering，2008，35（7）：90－94．
[94] D．Gao，Y．X．Liu，X．G．Zhang，Y．X．Liu．Binary-tree multi-classifier for weldingdefects and its application based on SVM [C]．Proceeding of the8th World Congress onIntelligent Control and Automation，2006：8509-8513．
[95]孙林，杨世元，吴德会．X射线底片焊缝缺陷的支持向量机识别方法[J]．应用科学学报，2008，26（4）：418－424．
[96] H．G．Bu，J．Wang，X．B．Huang．Fabric defect detection based on multiple fractal featuresand support vector data description [J]．Engineering Applications of Artificial Intelligence，2009，22：224－235．
[97] T．S．Li．Applying wavelets transforms，rough set theory and support vector machine forcopper clad laminate defects classification [J]．Exports Systems with Applications，2009，36：5822－5829．
[98]贺秋伟，王龙山，于忠党，李国发，高立国．基于图像处理和支持向量机的微型齿轮缺陷检测[J]．吉林大学学报，2008，38（5）：565－569．
[99] T．W．Liao，A．K．Celmins，R．J．Hammell Ⅱ．A fuzzy c-means variant for the generationof fuzzy term sets [J]．Fuzzy Sets and System，2003，135：241－257．
[100] T．W．Liao．Classification of welding flaw types with fuzzy expert systems [J]．ExpertSystems with Application，2003，25：101－111．
[101]李永生，戚大伟．基于FCM的木材X射线图像缺陷分割[J]．森林工程，2008，24（3）：51－52．
[102] F．Herold，K．Bavendiek，R．Grigat．A Third Generation Automatic Defect Detection System[R]．
[103] L．Qin，Q．F．Zheng，et al．Unsupervised texture classification：Automatically discoverand classify texture patterns [J]．Image and Vision Computing，2008，26：647－656．
[104] Y．L．Qiao，Z．M．Lu，J．S．Pan，S．H．Sun．Fast k-nearest neighbor search algorithmbased on pyramid structure of wavelet transform and its application to texture classification[J]．Digital Signal Processing：A Review Journal，2010，20（3）：837－845．
[105] D．Le and S．Satoh．Ent-Boost：Boosting using entropy measures for robust object detection[J]．Pattern Recognition Letters，2007，28：1083－1090．
[106]林开颜，吴军辉．基于计算机视觉的水果分级技术研究进展[J]．信息化纵横，2009，（10）：1－9．
[107]陈小娜，章程辉．绿橙表面缺陷的计算机视觉分级技术[J]．农机化研究，2009，（10）：126－129．
[108]龚毅光，白俊杰，王宁生．汽车产品审核中智能评价缺陷技术的研究与应用[J]．机械科学与技术，2008，27（12）：1552－1558．
[109]孙朝明，汤光平，李强，王增勇．数字化射线检测图像的缺陷评级方法[C]．全国射线检测技术年会论文集，2011：75－81．
[110]徐兴林，贾国华．双臂双投影法透照椭圆一次成像在小径管焊口缺陷评级中的使用[J]．热力发电，2005，（6）：50－54．
[111]庄天戈．CT理论与算法[M]．上海：上海交通大学出版社，1992．
[112] G．L．Zeng．Medical Image Reconstruction [M]．Springer，2009．
[113]叶海霞．工业CT窄角扇束卷积反投影并行图像重建研究[D]．重庆：重庆大学，2003．
[114]叶云长．计算机层析成像检测[M]．北京：机械工业出版社，2006．
[115] http://www.volren.org/[EB/OL]．
[116] M．Q．Yang，Y．H．Peng，Y．X．Liu．The algorithm and application of finite line integraltransform[C]．IEEE International Symposium on Microwave，Antenna，Propagation and EMCTechnologies Proceedings，2005：411－414．
[117] K．I．Laws．Texture image segmentation [R]．Image Processing Institute，Univ. of SouthernCalifornia，1980．
[118] S．Q．Zhao，Y．S．Gao，B．C．Zhang．Sobel-LBP [C]．In：IEEE Internat．Conf．onImage Proces．（ICIP），2008：2144－2147．
[119] Q．Wang，J．M．Liang，et al．．Yang．Spatial texture based automatic classification of daysideaurora in all-sky image [J]．J．Atmos．Sol．-Terr．Phys．，2010，72：498－508．
[120] M．M．Azab，H．A．Shedeed，A．S．Hussein．A new technique for background modelingand subtraction for motion detection in real-time videos [C]．IEEE Internat．Conf．on ImageProces．（ICIP），2010：3453－3456．
[121] H．Sun，C．Wang，B．L．Wang，N．EL-Sheimy．Pyramid binary pattern features for real-timepedestrian detection from infrared videos [J]．Neurocomputing，2011，74：797－804．
[122] K．B．Wang，B．Z．Yu，W．Xi．Radar image segmentation using active contour [C]．20071st Asian and Pacific Conference on Synthetic Aperture Radar Pages，2007：446－450．
[123] Y．Li，S．Luo，Q．Zou．Learning to detect boundaries in natural image using texture cuesand EM [C]．2008Fourth International Conference on Natural Computation（ICNC），2008：167－171．
[124] M．A．Savelonas，D．K．Iakovidis，D．Maroulis．LBP－guided active contours [J]．PatternRecognit．Lett．，2008，29：1404－1415．
[125]曾理，蒲云，马睿．基于工业CT的铁路货车铸件缺陷自动检测[J]．中国铁道科学，2009，30（4）：76－80．
[126] H．L．Jin，Q．S．Liu，et al．．Face detection using improved LBP under Bayesian framework
[C]．Proceedings. Third International Conference on Image and Graphics，2004：306－309｜xvii+588．
[127] Y．Wang，X．Y．Wei，S．Xiao．LBP texture analysis based on the local adaptive Niblackalgorithm [C]．CISP2008：Proceedings of First International Congress on Image and SignalProcessing，2008，2：777－780．
[128] G．D．Tian and A．D．Men．An improved texture-based method for background subtractionusing local binary patterns [C]．Proceedings of The20092ndInternational Congress on Imageand Signal Processing，2009，1－9：1984－1987．
[129] http://www.3dmed.net/[EB/OL]．
[130] L．Wang，L．He，A．Mishra，C．M．Li．Active contours driven by local Gaussian distributionfitting energy [J]．Signal Process，2009，89：2435－2447．