基于机器视觉的贴片安装锡膏三维检测
|
摘要
电子制造业的印刷电路板上元器件安装采用表面贴片安装技术(SMT, Surface Mounted Technology)后,元器件密度、生产自动化水平、产品质量都得到很大提高。但检测技术的不足却影响了产品质量和生产自动化水平。而引起电子产品质量缺陷的一个重要原因就是贴片安装时锡膏印刷的缺陷。锡膏印刷质量检测包括对所印刷锡膏面积、位置的二维检测,还包括对锡膏厚度、体积和形状的三维测量。传统的借助放大镜的人工目检无法适应高集成度、细间距、微小元件的锡膏检测。采用激光三角法半自动锡膏检测也无法满足贴片安装生产的精度和速度要求。目前,国内贴片安装生产线使用的锡膏三维测试仪器完全依靠进口,设备的价格非常昂贵。因此,研发具有自主知识产权的三维锡膏检测设备,对于提高电子制造业自动化水平、降低生产成本、提高电子产品质量都具有重要意义。
三维测量的方法有很多,传统的机械接触式测量法(如探针式),由于存在测量力、测量时间长、需进行测头半径的补偿、不能测量较软质材料等局限性,为此人们寻求能够克服上述缺陷的三维测量方法,非接触的光学投影式三维轮廓术正是在这一要求下出现的。光学投影式三维轮廓术有多种,其中,相位法三维测量轮廓术,简称相位测量轮廓术,是属于投影式光学投影式三维轮廓术的一种,它以其非接触、高速度、高精度、大数据量等一系列优点而日益受到人们的重视和研究,另外该方法还具有受环境电磁场影响小、工作距离大、可测量非金属面及较软质材料等特点。
相位测量轮廓术根据观察物体上结构光的灰度被调制变化情况来测量物体的三维形貌。它通过多幅相移条纹图来计算得到截断的相位,再通过相位展开算法将截断相位展开为连续相位,最后根据物体高度对结构光相位的调制关系从计算得到的相位中恢复物体的表面三维高度。本论文的工作是在已有的研究工作基础上,重点研究了相位测量轮廓术中的条纹投影及获取,相位展开,相位-高度映射关系,测量系统标定等关键技术,分析了影响相位测量轮廓术测量精度的因素。
用结构光投形进行机器视觉三维测量可以快速有效检测电子产品贴片生产线锡膏印刷质量。图像采集系统的设计其中主要环节,通过选取合适的摄像机、镜头和投影仪,建立了三维测量系统。通过标定板,完成对图像采集系统的准确标定。实验结果初步验证了设计的图像采集系统经标定后可以满足三维检测的精度要求。本文还详细分析了相位展开算法的基本原理,讨论和比较了各种算法的设计方法、计算公式和适用范围,对相位展开方法的最新进展和应用前景进行了较全面的分析和评述。
After using surface Mounted Technology (SMT) installed components, the components density, production automation level, product quality of the electronic manufacturing printed circuit boards are improved. But testing technology deficiency badly affects the quality of products and production automatization level. An important reason of Electronic product quality defects is the solder paste printing defects when Surface Mounted. The printing quality testing including the two-dimensional detection of the area、position and the 3d measurement of the thickness、volume and shapes of solder paste printing. The traditional using a magnifying glass artificial visual inspection can't adapt the solder paste testing of high level of integration, fine spacing, tiny components. By using the laser triangulation semi-auto solder paste test can't satisfy the patch installation precision and speed production requirements. At present, the domestic PCB (Printed Circle Board) installation of the line completely dependent on imports of 3d test instrument.3D measurement systems of solder paste completely rely on imports, But the imported equipments are more expensive. So that, developing 3D measurement systems of solder paste with independent intellectual property to enhance electronic manufacturing automation level, reduce the production cost, improve electronic product quality is very significant.
There are a lot of methods for 3D shape measurement. Because the traditional mechanical contact profilometry has many limitations, for example, the measurement speed is low, it exists measuring power and need compensate the size of the probe, it can't measure the soft material. Therefor, the People seek a better kind of profilometry which is able to overcome these limitations, untouched optical profilometry based on light pattern Projection is such an profilometry. Untouched optical profilometry based on light pattern projection has many types. One is grating projection phase measurement method. It has many advantages such as untouched, high speed, high precision, more data. It also isn't affected by the electromagnetic wave in the surrounding environment. It can measure the nonmetal object and soft material. Phase measurement method is becoming more and more important for his advantages.
3D shape of objects is measured by observation the change of structure light gray of the object contour based on objects in PMP.The wrapped phase of fringe pattern is calculated from more than three frames of images containing phase shifted fringe pattern. Then, the continuous phase is obtained through unwrapping algorithm. Finally, Finally,according to the objects height on the structure light phase modulation relationship from the calculated phase to recover the 3d surface height objects. In this dissertation, the key technologies in PMP such as casting and acquiring fringe pattern, phase unwrapping, Phase-height mapping and system calibrating etc. The factors affecting measurement accuracy in PMP are analyzed.
Structured light based machine vision three-dimensional measurement can quickly and effectively inspect solder paste defaults in SMT assembly line. Image acquisition designing is an important part of the system, which was composed of proper selected camera, suited lens and sinusoidal grating projector. Image acquisition subsystem calibration is also necessary and was proposed by standard calibration block. Experiment results verify the designed of image collection system can satisfy the calibration of the 3D measurement precision requirements. In the dissertation we also analyzed the basic principle of the phase unwrapping algorithm in detail, and we also discussed and compared the designing method, formula and applying range of each algorithm. At last, we analyzed and gave comments on the new development and its application.
三维测量的方法有很多,传统的机械接触式测量法(如探针式),由于存在测量力、测量时间长、需进行测头半径的补偿、不能测量较软质材料等局限性,为此人们寻求能够克服上述缺陷的三维测量方法,非接触的光学投影式三维轮廓术正是在这一要求下出现的。光学投影式三维轮廓术有多种,其中,相位法三维测量轮廓术,简称相位测量轮廓术,是属于投影式光学投影式三维轮廓术的一种,它以其非接触、高速度、高精度、大数据量等一系列优点而日益受到人们的重视和研究,另外该方法还具有受环境电磁场影响小、工作距离大、可测量非金属面及较软质材料等特点。
相位测量轮廓术根据观察物体上结构光的灰度被调制变化情况来测量物体的三维形貌。它通过多幅相移条纹图来计算得到截断的相位,再通过相位展开算法将截断相位展开为连续相位,最后根据物体高度对结构光相位的调制关系从计算得到的相位中恢复物体的表面三维高度。本论文的工作是在已有的研究工作基础上,重点研究了相位测量轮廓术中的条纹投影及获取,相位展开,相位-高度映射关系,测量系统标定等关键技术,分析了影响相位测量轮廓术测量精度的因素。
用结构光投形进行机器视觉三维测量可以快速有效检测电子产品贴片生产线锡膏印刷质量。图像采集系统的设计其中主要环节,通过选取合适的摄像机、镜头和投影仪,建立了三维测量系统。通过标定板,完成对图像采集系统的准确标定。实验结果初步验证了设计的图像采集系统经标定后可以满足三维检测的精度要求。本文还详细分析了相位展开算法的基本原理,讨论和比较了各种算法的设计方法、计算公式和适用范围,对相位展开方法的最新进展和应用前景进行了较全面的分析和评述。
After using surface Mounted Technology (SMT) installed components, the components density, production automation level, product quality of the electronic manufacturing printed circuit boards are improved. But testing technology deficiency badly affects the quality of products and production automatization level. An important reason of Electronic product quality defects is the solder paste printing defects when Surface Mounted. The printing quality testing including the two-dimensional detection of the area、position and the 3d measurement of the thickness、volume and shapes of solder paste printing. The traditional using a magnifying glass artificial visual inspection can't adapt the solder paste testing of high level of integration, fine spacing, tiny components. By using the laser triangulation semi-auto solder paste test can't satisfy the patch installation precision and speed production requirements. At present, the domestic PCB (Printed Circle Board) installation of the line completely dependent on imports of 3d test instrument.3D measurement systems of solder paste completely rely on imports, But the imported equipments are more expensive. So that, developing 3D measurement systems of solder paste with independent intellectual property to enhance electronic manufacturing automation level, reduce the production cost, improve electronic product quality is very significant.
There are a lot of methods for 3D shape measurement. Because the traditional mechanical contact profilometry has many limitations, for example, the measurement speed is low, it exists measuring power and need compensate the size of the probe, it can't measure the soft material. Therefor, the People seek a better kind of profilometry which is able to overcome these limitations, untouched optical profilometry based on light pattern Projection is such an profilometry. Untouched optical profilometry based on light pattern projection has many types. One is grating projection phase measurement method. It has many advantages such as untouched, high speed, high precision, more data. It also isn't affected by the electromagnetic wave in the surrounding environment. It can measure the nonmetal object and soft material. Phase measurement method is becoming more and more important for his advantages.
3D shape of objects is measured by observation the change of structure light gray of the object contour based on objects in PMP.The wrapped phase of fringe pattern is calculated from more than three frames of images containing phase shifted fringe pattern. Then, the continuous phase is obtained through unwrapping algorithm. Finally, Finally,according to the objects height on the structure light phase modulation relationship from the calculated phase to recover the 3d surface height objects. In this dissertation, the key technologies in PMP such as casting and acquiring fringe pattern, phase unwrapping, Phase-height mapping and system calibrating etc. The factors affecting measurement accuracy in PMP are analyzed.
Structured light based machine vision three-dimensional measurement can quickly and effectively inspect solder paste defaults in SMT assembly line. Image acquisition designing is an important part of the system, which was composed of proper selected camera, suited lens and sinusoidal grating projector. Image acquisition subsystem calibration is also necessary and was proposed by standard calibration block. Experiment results verify the designed of image collection system can satisfy the calibration of the 3D measurement precision requirements. In the dissertation we also analyzed the basic principle of the phase unwrapping algorithm in detail, and we also discussed and compared the designing method, formula and applying range of each algorithm. At last, we analyzed and gave comments on the new development and its application.
引文
[1]罗兵,章云.基于深度图像的SMT锡膏印刷实时自动检测[J].无损检测,2007,29(9)：519-525.
[2]刘斌.锡膏三维测量系统的研究[D].天津大学硕士学位论文,2007.
[3]张永忠.3D锡膏测量仪的研究[D].哈尔滨工业大学硕士学位论文,2006.
[4]Huihui Wu, Xianmin Zhang, Yongcong Kuang, Shenglin Lu. A Real-time Machine Vision Snystem for Solder Paste Inspection [A]. Proc. of IEEE/ASME Internatioal Conference on Advanced Intelligent Mechatronics[C],2008:14(7):205-210.
[5]王鹏.线结构光三维自动扫描系统关键技术的研究[D].天津大学博士学位论文,2008.
[6]吴兆华,周德俭.电路模块表面组装技术[M].北京人民邮电出版社,2008,7.
[7]国家自然科学基金项目查询[ED/OL]. http://159.226.244.28/portal/proj_search. asp/,2009.
[8]Xinyu Wu, Wingkwong Chung, Hang Tong, Jun Cheng and Yangsheng Xu. A New Solder Paste Inspection Device:Design and Algorithm [A]. Proc. of IEEE International Conference on Robotics and Automation[C],2007:680-685.
[9]常素萍.基于白光干涉轮廓尺寸与形貌非接触测量方法和系统[D].华中科技大学博士学位论文,2007.
[10]李满海.傅立叶变换轮廓术的若干问题研究[D].四川大学硕士学位论文,2004.
[11]李国平.基于莫尔技术的标牌凹凸字符图像获取与识别研究[D].山东大学博士学位论文,2007.
[12]孙姣芬.虚光栅移相莫尔条纹法对干涉图的处理及应用[D].南京理工大学硕士学位论文,2004.
[13]刘世峰.基于幅值采样的光栅莫尔条纹信号细分技术的研究[D].华中科技大学硕士学位论文,2007.
[14]李健兵.基于投影条纹的物体三维形貌精密测量方法的研究[D].国防科学技术大学硕士学位论文,2004.
[15]段存丽,陈芳,祁瑞利,张苏娟.利用莫尔条纹测量物体三维形貌新方法研究[J].光子学报,2008,37(7)：1425-1428.
[16]Song Zhang, PeisenS.Huang. High-resolution, real-time three-dimensional shape measurement [J]. Optical Engineering,2006,45(12):123601(1-8).
[17]Xinyu Wu, Wingkwong Chung, Jun Cheng, Hang Tong and Yangsheng Xu. A Parallel-Structure Solder Paste Inspection System [A]. Proc. of IEEE/ASME Transactions On Mechatronics[C],2009,(14):590-597.
[18]赵焕东.相位测量轮廓术的理论研究及应用[D].浙江大学博士学位论文,2001.
[19]牛小兵.光栅投射三维轮廓测量技术的研究[D].天津大学博士学位论文,2003.
[20]李勇.相位测量轮廓术关键技术及应用研究[D].四川大学博士学位论文,2006.
[21]Grantham K.H.Pang, Ming-Hei Chu. Automated Optical Inspection of Solder Paste based on 2.5D Visual Images [A]. Proc. of IEEE International Conference on Mechatronics and Automation[C],2009,12(8).
[22]刘玉凤,曹益平,何宇航.新改进型三步非等步相位测量轮廓术算法[J].光电子激光,2010,21(1)：83-86.
[23]李中伟.基于数字光栅投影的结构光三维测量技术与系统研究[D].华中科技大学博士学位论文,2009.
[24]吴双卿.光栅投影三维形貌测量技术的研究[D].西南交通大学硕士学位论文,2005.
[25]罗兵,章云.SMT锡膏印刷质量自动光学检测[J].电子质量,2005,(12)：30-32.
[26]Luo Bing, Zhang Li-Yun. SMT solder paste deposit inspection based on 3D PMP and 2D image features fusion [A]. Proc. of IEEE Int. Conf. on Wavelet Analysis and Pattern Recognition [C],2010:190-194.
[27]邱崧.基于LED光源的DLP投影系统研究[D].华东师范大学博士学位论文：2007.
[28]杨凤和.DLP技术及其投影显示应用研究[D].中国科学院研究生院博士学位论文,2005.
[29]Steger C, Ulrish M, Wiedemann C. Machine Vision Algorithms and Applications [M].北京：清华大学出版社,2008.11.
[30]张广军.视觉测量[M].北京：科学出版社,2008,3.
[31]Zhang Zheng-you. A flexible new technique for camera calibration [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2000,22(11): 1330-1334.
[32]Zhang S, Huang P S, Phase error compensation for a 3D shape measurement system based on the phase-shifting method[J]. Optical Engineering,2007,46(6):1-9.
[33]Legarda-Saenz R, Bothe T, Juptner W P. Accurate procedure for the calibration of a structured light system [J]. Optical Engineering,2004,43(2):464-471.
[34]韦争亮,钟约先,袁朝龙等.单摄像机单投影仪三维测量系统标定技术[J].清华大学学报(自然科学版),2009,49(2)：202-205.
[35]张贵姣.基于多频外差原理的相位测量关键技术研究[D].华中科技大学硕士学位论文,2007.
[36]薛莲.基于条纹频率分析的位相展开新算法[D].四川大学硕士学位论文,2001.
[37]杨锋涛,罗江龙,刘志强,吕晓旭.相位展开的6种算法比较[J].激光技术2008,32(3)：323-326.
[38]杨锋涛.相位展开算法及在相位测量轮廓术中的应用研究[D].昆明理工大学硕士学位论文,2005.
[39]岳慧敏.基于时间相位展开的三维轮廓测量研究[D].四川大学博士学位论文,2005.
[40]刘景峰,李艳秋,刘克.含噪声包裹相位图的加权最小二乘相位展开算法研究[J].光学技术,2008,34(5)：643-650.
[41]朱越.光学相位测量中余弦相位展开方法研究[D].昆明理工大学硕士学位论文,2004.
[42]葛晶许,增朴,于德敏等.三维物体表面轮廓测量中实用相位展开算法的研究[J].微计算机信息,2006,22(10-3)：290-292.
[43]Song Zhang, Peisen S. Huang. Phase error compensation for a 3-D shape measurement system based on the phase-shifting method[J]. Optical Engineering, 2007,46(6):063601(1-9).
[44]谢晶,贾克斌.一种基于二维直方图的H.264/AVC快速帧内预测判决算法[J].电子与信息学报,2005,27(7)：1053-1057.
[45]Lin Yu-Kun, Chang Tian-Sheuan. Fast block type decision algorithm for intra prediction in H.264 FRE×t[A]. Proc. of IEEE International Conference on Image Processing[C], Genova,2005:585-588.
[46]Meng Qing-lei, Yao Chun-lian, Song Jian-bin, Li Wei. Fast selective algorithm of Intra prediction for H.264/AVC[J]. Journal of Beijing University of Aeronautics and Astronautics,2007,33(2):219-223.
[47]JONGHO KIM, JECHANG JEONG Fast intra-mode decision in H.264 video coding using simple directional masks1[A]. Proceedings of SPIE[C].2005,1071-1079.
[48]王德宝.基于H.264的帧内与帧间预测方法研究[D].合肥工业大学硕士学位论文,2007.
[49]毕厚杰.新一代视频编码压缩标准—H.264/AVC[M].北京：人民邮电出版社,2005,4.
[50]张春田,苏育挺,张静.数字图像压缩编码[M].北京：清华大学出版社,2006,09.
[51]李秋山,马妍.H.264帧内预测模式的快速选择算法[J].计算机工程与设计.2009,30(22)：5136-5139.
[52]谢翠兰,郑艺玲.一种基于H.264的帧内模式选择快速算法[J].广西工学院学报,2008,19(2)：66-70.
[2]刘斌.锡膏三维测量系统的研究[D].天津大学硕士学位论文,2007.
[3]张永忠.3D锡膏测量仪的研究[D].哈尔滨工业大学硕士学位论文,2006.
[4]Huihui Wu, Xianmin Zhang, Yongcong Kuang, Shenglin Lu. A Real-time Machine Vision Snystem for Solder Paste Inspection [A]. Proc. of IEEE/ASME Internatioal Conference on Advanced Intelligent Mechatronics[C],2008:14(7):205-210.
[5]王鹏.线结构光三维自动扫描系统关键技术的研究[D].天津大学博士学位论文,2008.
[6]吴兆华,周德俭.电路模块表面组装技术[M].北京人民邮电出版社,2008,7.
[7]国家自然科学基金项目查询[ED/OL]. http://159.226.244.28/portal/proj_search. asp/,2009.
[8]Xinyu Wu, Wingkwong Chung, Hang Tong, Jun Cheng and Yangsheng Xu. A New Solder Paste Inspection Device:Design and Algorithm [A]. Proc. of IEEE International Conference on Robotics and Automation[C],2007:680-685.
[9]常素萍.基于白光干涉轮廓尺寸与形貌非接触测量方法和系统[D].华中科技大学博士学位论文,2007.
[10]李满海.傅立叶变换轮廓术的若干问题研究[D].四川大学硕士学位论文,2004.
[11]李国平.基于莫尔技术的标牌凹凸字符图像获取与识别研究[D].山东大学博士学位论文,2007.
[12]孙姣芬.虚光栅移相莫尔条纹法对干涉图的处理及应用[D].南京理工大学硕士学位论文,2004.
[13]刘世峰.基于幅值采样的光栅莫尔条纹信号细分技术的研究[D].华中科技大学硕士学位论文,2007.
[14]李健兵.基于投影条纹的物体三维形貌精密测量方法的研究[D].国防科学技术大学硕士学位论文,2004.
[15]段存丽,陈芳,祁瑞利,张苏娟.利用莫尔条纹测量物体三维形貌新方法研究[J].光子学报,2008,37(7)：1425-1428.
[16]Song Zhang, PeisenS.Huang. High-resolution, real-time three-dimensional shape measurement [J]. Optical Engineering,2006,45(12):123601(1-8).
[17]Xinyu Wu, Wingkwong Chung, Jun Cheng, Hang Tong and Yangsheng Xu. A Parallel-Structure Solder Paste Inspection System [A]. Proc. of IEEE/ASME Transactions On Mechatronics[C],2009,(14):590-597.
[18]赵焕东.相位测量轮廓术的理论研究及应用[D].浙江大学博士学位论文,2001.
[19]牛小兵.光栅投射三维轮廓测量技术的研究[D].天津大学博士学位论文,2003.
[20]李勇.相位测量轮廓术关键技术及应用研究[D].四川大学博士学位论文,2006.
[21]Grantham K.H.Pang, Ming-Hei Chu. Automated Optical Inspection of Solder Paste based on 2.5D Visual Images [A]. Proc. of IEEE International Conference on Mechatronics and Automation[C],2009,12(8).
[22]刘玉凤,曹益平,何宇航.新改进型三步非等步相位测量轮廓术算法[J].光电子激光,2010,21(1)：83-86.
[23]李中伟.基于数字光栅投影的结构光三维测量技术与系统研究[D].华中科技大学博士学位论文,2009.
[24]吴双卿.光栅投影三维形貌测量技术的研究[D].西南交通大学硕士学位论文,2005.
[25]罗兵,章云.SMT锡膏印刷质量自动光学检测[J].电子质量,2005,(12)：30-32.
[26]Luo Bing, Zhang Li-Yun. SMT solder paste deposit inspection based on 3D PMP and 2D image features fusion [A]. Proc. of IEEE Int. Conf. on Wavelet Analysis and Pattern Recognition [C],2010:190-194.
[27]邱崧.基于LED光源的DLP投影系统研究[D].华东师范大学博士学位论文：2007.
[28]杨凤和.DLP技术及其投影显示应用研究[D].中国科学院研究生院博士学位论文,2005.
[29]Steger C, Ulrish M, Wiedemann C. Machine Vision Algorithms and Applications [M].北京：清华大学出版社,2008.11.
[30]张广军.视觉测量[M].北京：科学出版社,2008,3.
[31]Zhang Zheng-you. A flexible new technique for camera calibration [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2000,22(11): 1330-1334.
[32]Zhang S, Huang P S, Phase error compensation for a 3D shape measurement system based on the phase-shifting method[J]. Optical Engineering,2007,46(6):1-9.
[33]Legarda-Saenz R, Bothe T, Juptner W P. Accurate procedure for the calibration of a structured light system [J]. Optical Engineering,2004,43(2):464-471.
[34]韦争亮,钟约先,袁朝龙等.单摄像机单投影仪三维测量系统标定技术[J].清华大学学报(自然科学版),2009,49(2)：202-205.
[35]张贵姣.基于多频外差原理的相位测量关键技术研究[D].华中科技大学硕士学位论文,2007.
[36]薛莲.基于条纹频率分析的位相展开新算法[D].四川大学硕士学位论文,2001.
[37]杨锋涛,罗江龙,刘志强,吕晓旭.相位展开的6种算法比较[J].激光技术2008,32(3)：323-326.
[38]杨锋涛.相位展开算法及在相位测量轮廓术中的应用研究[D].昆明理工大学硕士学位论文,2005.
[39]岳慧敏.基于时间相位展开的三维轮廓测量研究[D].四川大学博士学位论文,2005.
[40]刘景峰,李艳秋,刘克.含噪声包裹相位图的加权最小二乘相位展开算法研究[J].光学技术,2008,34(5)：643-650.
[41]朱越.光学相位测量中余弦相位展开方法研究[D].昆明理工大学硕士学位论文,2004.
[42]葛晶许,增朴,于德敏等.三维物体表面轮廓测量中实用相位展开算法的研究[J].微计算机信息,2006,22(10-3)：290-292.
[43]Song Zhang, Peisen S. Huang. Phase error compensation for a 3-D shape measurement system based on the phase-shifting method[J]. Optical Engineering, 2007,46(6):063601(1-9).
[44]谢晶,贾克斌.一种基于二维直方图的H.264/AVC快速帧内预测判决算法[J].电子与信息学报,2005,27(7)：1053-1057.
[45]Lin Yu-Kun, Chang Tian-Sheuan. Fast block type decision algorithm for intra prediction in H.264 FRE×t[A]. Proc. of IEEE International Conference on Image Processing[C], Genova,2005:585-588.
[46]Meng Qing-lei, Yao Chun-lian, Song Jian-bin, Li Wei. Fast selective algorithm of Intra prediction for H.264/AVC[J]. Journal of Beijing University of Aeronautics and Astronautics,2007,33(2):219-223.
[47]JONGHO KIM, JECHANG JEONG Fast intra-mode decision in H.264 video coding using simple directional masks1[A]. Proceedings of SPIE[C].2005,1071-1079.
[48]王德宝.基于H.264的帧内与帧间预测方法研究[D].合肥工业大学硕士学位论文,2007.
[49]毕厚杰.新一代视频编码压缩标准—H.264/AVC[M].北京：人民邮电出版社,2005,4.
[50]张春田,苏育挺,张静.数字图像压缩编码[M].北京：清华大学出版社,2006,09.
[51]李秋山,马妍.H.264帧内预测模式的快速选择算法[J].计算机工程与设计.2009,30(22)：5136-5139.
[52]谢翠兰,郑艺玲.一种基于H.264的帧内模式选择快速算法[J].广西工学院学报,2008,19(2)：66-70.