基于光学小波变换的机器人视觉系统及其实验研究
|
摘要
机器人视觉技术是80年代发展起来的一门新兴技术，是当今国际研究的前沿课题。近几年，机器人视觉技术取得了一定的进展，并出现了一些成功的机器人视觉系统，但是由于它是一门交叉性很强的学科，研究工作碰到了相当多的问题，总的来说直到今天还没有一个成功的通用系统出现，理论体系还处于一个百花齐放、百家争鸣的孕育期。机器人视觉系统有其自身的局限性，实现起来还存在很多问题。因为机器人视觉技术的特点之一就是数据量庞大，处理这些数据需要很长的时间，这样，处理速度就成为机器人视觉系统从实用角度急待解决的一个瓶颈问题。目前解决的方法主要是用专用的DSP芯片，但其通用性不高，设计成本昂贵。因此许多科技人员开始从光学信息处理技术寻求提高机器人视觉处理速度的途径。
光学信息处理是近几年来发展起来的一门新兴学科，指的是用光学方法实现对输入信息的各种变换或处理。与其它形态的信号处理相比，光学信息处理系统具有处理速度快、容量大、特别适合处理图象等优势，其中研究的前沿领域就包括光学小波变换。小波变换(Wavelet Transform)是80年代后期发展起来的应用数学分支，它起源于信号分析领域，被认为是近十几年来在数学工具和方法上的重大突破。光学小波变换(Optical Wavelet Transform)是基于小波变换的基本理论利用光学技术与方法对信号与图象的小波变换处理。由于光学小波变换同时具有光学信息处理系统高度并行、大容量的特点和小波变换特别适合于处理图象信号的特点，因此目前它已广泛应用于图象特征提取、目标探测与识别等领域。
正是基于光学小波变换在图象处理领域的巨大优势，本文创造性的将其引入机器人视觉，并设计出一套基于光学小波变换的机器人视觉系统，同时对光学小波变换滤波器的设计进行了深入研究，并通过编程的方法实现了小波滤波器的构造，克服了目前光学小波变换灵活性不高，可操作性不强等缺陷，同时对二维小波变换对于图象边缘提取的有时进行了理论分析。本文设计出用于半导体激光器的扩束准直机构，并成功获得实验所需的准直相干光，本文还成功完成了机器人视觉系统目标读取部分的实验。进一步从实验角度证明了该机器人视觉系统的可行性。
Robot vision is a new technology developed since 1980s. It is also the forward issues all over the world. In recent years, robot vision has developed fast and some succeeded robot vision system had been designed. But because it is a complex technology and there are many difficulties in research, there is not a succeeded versatile vision system yet. Robot vision system has many limitations, one of which is its data processing are very large so that it has to spend much time to process these data. Therefore, the processing speed becomes one of the vital problems. At present time the method to improve the processing speed is to use personal DSP chips. But the DSP chips are not versatile and cost much, so more and more technologists start to improve robot vision processing speed with optical information processing.
Optical information processing is a novel technology developed recent years. It realizes transforms or processing of the input information by optical method. Its advantages include high processing speed, large storage, and are highly suit to process image information compared to other signal processing methods. One of its forward research issues is optical wavelet transform. Wavelet transform is a branch of applied mathematics. It derives from signal analysis is considered the important breakthrough of applied mathematics in recent decades. Optical wavelet transform realize wavelet transform of signal and image through optical technology and method-based wavelet transform theory. Today optical wavelet transform has been used in image edge extraction, object detection and recognition etc because it has not only the advantages of optical information processing system but also the advantages of wavelet transform.
It is because optical wavelet transform has great advantages in image processing field that this paper use optical wavelet transform in robot vision creatively and design a novel robot vision system based on optical wavelet transform. And this paper also researched the optical wavelet transform filter deeply and realize a novel wavelet filter by programming method which overcome the defects of
current wavelet filter, such as low flexibility and operability. The theory analysis of 2-D wavelet transforms utility in image edge extraction is also gives in this paper. A collimating device is also designed in this paper, and the collimating coherent light is good proved by the followed experiments. The robot vision system's object reading experiment is also finished in this paper, all these proved that this novel robot vision system is feasible.
光学信息处理是近几年来发展起来的一门新兴学科，指的是用光学方法实现对输入信息的各种变换或处理。与其它形态的信号处理相比，光学信息处理系统具有处理速度快、容量大、特别适合处理图象等优势，其中研究的前沿领域就包括光学小波变换。小波变换(Wavelet Transform)是80年代后期发展起来的应用数学分支，它起源于信号分析领域，被认为是近十几年来在数学工具和方法上的重大突破。光学小波变换(Optical Wavelet Transform)是基于小波变换的基本理论利用光学技术与方法对信号与图象的小波变换处理。由于光学小波变换同时具有光学信息处理系统高度并行、大容量的特点和小波变换特别适合于处理图象信号的特点，因此目前它已广泛应用于图象特征提取、目标探测与识别等领域。
正是基于光学小波变换在图象处理领域的巨大优势，本文创造性的将其引入机器人视觉，并设计出一套基于光学小波变换的机器人视觉系统，同时对光学小波变换滤波器的设计进行了深入研究，并通过编程的方法实现了小波滤波器的构造，克服了目前光学小波变换灵活性不高，可操作性不强等缺陷，同时对二维小波变换对于图象边缘提取的有时进行了理论分析。本文设计出用于半导体激光器的扩束准直机构，并成功获得实验所需的准直相干光，本文还成功完成了机器人视觉系统目标读取部分的实验。进一步从实验角度证明了该机器人视觉系统的可行性。
Robot vision is a new technology developed since 1980s. It is also the forward issues all over the world. In recent years, robot vision has developed fast and some succeeded robot vision system had been designed. But because it is a complex technology and there are many difficulties in research, there is not a succeeded versatile vision system yet. Robot vision system has many limitations, one of which is its data processing are very large so that it has to spend much time to process these data. Therefore, the processing speed becomes one of the vital problems. At present time the method to improve the processing speed is to use personal DSP chips. But the DSP chips are not versatile and cost much, so more and more technologists start to improve robot vision processing speed with optical information processing.
Optical information processing is a novel technology developed recent years. It realizes transforms or processing of the input information by optical method. Its advantages include high processing speed, large storage, and are highly suit to process image information compared to other signal processing methods. One of its forward research issues is optical wavelet transform. Wavelet transform is a branch of applied mathematics. It derives from signal analysis is considered the important breakthrough of applied mathematics in recent decades. Optical wavelet transform realize wavelet transform of signal and image through optical technology and method-based wavelet transform theory. Today optical wavelet transform has been used in image edge extraction, object detection and recognition etc because it has not only the advantages of optical information processing system but also the advantages of wavelet transform.
It is because optical wavelet transform has great advantages in image processing field that this paper use optical wavelet transform in robot vision creatively and design a novel robot vision system based on optical wavelet transform. And this paper also researched the optical wavelet transform filter deeply and realize a novel wavelet filter by programming method which overcome the defects of
current wavelet filter, such as low flexibility and operability. The theory analysis of 2-D wavelet transforms utility in image edge extraction is also gives in this paper. A collimating device is also designed in this paper, and the collimating coherent light is good proved by the followed experiments. The robot vision system's object reading experiment is also finished in this paper, all these proved that this novel robot vision system is feasible.
引文
[1] 王积分，张新荣．机器人视觉及其实例．机器人，1987，Vol．1(No．4)：56～60
[2] 傅京孙，蔡自兴，徐光佑．人工智能及其应用．北京：清华大学出版社，1987
[3] 钟玉琢，乔秉新，李树青．机器人视觉技术．北京：国防工业出社，1994
[4] 郑时雄，刘桂雄，阎华，等．机器人视觉在自动装配线上的应用．光学精密工程，1998，Vol 6(No．1)：90～94
[5] Bonarini, Andrea. Omnidirectioanl vision sensor for fast tracking for mobile robots. IEEE Transactions on Instrumentation and Measurement 49 3 Jun 2000 IEEE p 509-512 0018-9456
[6] 李桂成，卢长刚，王桂琴，等．一种消防机器人的视觉及其信号处理．机器人，1995，Vol．17，No．6．
[7] 王守杰，宗光华，毕树生．微操作机器人显微视觉系统的研究．机器人，1998，Vol．20 (No．6)：138～142
[8] 胡震，袁小海，陈容盛等．用于目标检测和精确定位的水下机器人视觉系统．中国造船，2000，Vol．41(No．2)：89～94
[9] Shimauchi K et al. Developmentof a hockey-robot using real-time vision, Proceedings of the 13th annual conference of RSJ., Kawasaki, 1995: 1137-1138
[10] 赵珠颖，佟国峰．机器人视觉子系统．机器人技术与应用，2000，No．2：26～28
[11] 邬永革，丁震，黄炯，等．一种有效的图象特征抽取和识别方法．计算机研究与发展，1997，Vol．34(No．1)：38～42
[12] 高速信息网络与并行处理研讨会论文集，国家自然科学基金委员会信息科学部，(1994．6)
[13] 清华大学光学仪器教研组．《信息光学基础》．北京：机械：工业出版社，1985
[14] A. Huang. Architectural consideration involved in the design of an optical digital computer..Proc. Of the IEEE, 1984, 72(7): 780-786
[15] A. Greenwood, M. Bartusiak, B. Burke et al. Science at the frontier. Volumel, National Academy Press, 1992
[16] A. W. Lohmann. What classical optics can do for the digital optical computer. Appl. Opt., 1986, 25(10): 1543-1549
[17] 吕乃光．傅立叶光学．北京：机械工业出版社，1988
[18] S. G. Mallat. Zero-Crossings of a wavelet transform. IEEE Trans. Inform. Theory, 1991, IT-37, 1019-1033
[19] O. Rioul and P. Duhamel. Fast algorithms for discrete and Continuous wavelet transforms. 1992, IEEE Trans. Inform. Theory, IT-38(2), 569-586
[20] Lee Jiann Shu, et al. Multiscale Corner Detection by Using Wavelet Transform. IEEE transaction on image processing, 1995, 14(1)
[21] 徐继友，朱耆祥．小波变换在图象边缘提取中的应用．光电工程，1998，Vol．25(No．4)：28～34
[22] 美国科学院，金国藩等．《科学前沿》．1993，第一卷：169～173
[23] 汤焱，莫玉龙．第二代小波变换应用于图象的无损压缩编码．中国图象图形学报，2000，Vol．5(A)(No．8)：699～702
[24] S. J. Schiff. Resolving time-series structure with a controlled Wavelet transform. Opt. Eng. 1992, Vol.31 (No.11): 2492
[25] 翁文国，廖光煊，范维澄．小波变换应用于DPIV图象去噪．中国图象图形学报，2000 Vol．5(A)(No．3)：211～215
[26] M. O. Freeman, A. Fedor, B. Bock et al. Optical Wavelet Processor for producing spatially localized ring-wedged-type information. SPIE Vo1.1772: 241-250(1992)
[27] 靳华，王晓丹，赵荣椿．基于小波变换和ODPSF的纹理图象分割方法．中国图象图形学报，2000 Vol．5(A)(No．10)：857～860
[28] D. Gabor. Theory of communication. J. Inst. Elect. Eng., 1946, Vol.93: 429-457
[29] J. M. Combes, A. Grossman, and Ph. Tchamitchin, eds. Wavelet, Time-Frequency Methods, and Phase Space(Springer-Verlag, New York, 1989): 68-98.
[30] A. Harr. Zur Theorie der orthogonalen Funktionen systeme. Math, Ann. 1910, 69: 331-371
[31] P. Burt and E. Adelson, "The Laplacian pyramid as a compact image code," IEEE Trans. Commun. 31, 532-540(1983)
[32] A. Grossman, J. Morlet. Decomposition of Hardy Function into square Integrable Wavelets of Constant Shape. SLAM. J. Math. Anal, 15(1984)
[33] S. Mallat. Multiresolution approximation and wavelets. Tech. Rep., Dep. Comput. Inform. Sci., Univ. Pennsylvania, Philadephia, PA. Sept(1987).
[34] H. Szu, Yunlong Sheng, and Jing Chen. Wavelet transform as a bank of the matched filters. Appl. Opt., 1992, 31(7): 3267-3277
[35] Guofan Jin, Yingbai Yan, Wenlu Wang et al. Optical Haar Wavelet Transform for Image Features Extraction. Proc. SPIE , 1993, Vol. 2034 Mathematical Imaging, San Diego.
[36] Special issue on Wavelet transform, Opt. Eng. 31, No.9, 1821---1916(1992).
[37] Yao li, and Yan Zhang. Coherent optical processing of Gabor and wavelet expansion of one and two-dimensional signals. Opt. Eng. , 1992, 31(9): 1865-1885
[38] Y. Meyer. In ondelettes et operatears. Paris:Hermann, 1990, pp30-36, 106-110
[39] 马晶，谭立英，冉启文．光学小波滤波理论初探．中国激光，1999，Vol．A26(No．4)：343～346
[40] X. Yang, H. Szu, Y. Sheng, and H. J. Caulfield, Opt. Eng., 31(9), 1846-1851(1992)
[41] T. Burns, K. Fielding, S. Rogers et al. Optical Haar wavelet transform. Opt. Eng., 1992,31(9): 1852-1858.
[42] 王文陆，严瑛白，金国藩，等．光学子波变换(2)实验结构与技术．光电子·激光，1994Vol．5(No．5)：269～272
[43] 康辉，许剑，战元龄，等．利用光折变全息术实现的光学小波变换．中国激光，1996，Vol．A23(No．8)：741～744
[44] 倪明，蒋志平，陆启生．一种光学小波变换的改进．中国激光，1997Vol．A24(No．3)：231～236
[45] 段发阶．计算机视觉检测基础理论及应用技术研究：[博士学位论文]．天津：天津大学，1994
[46] Roberts, L. G.(1965), Machine perception of 3D solids, J.T. Tippett et al eds., Optical and Electrooptical Information Processing, USA: MIT Press: 159-197
[47] Prewitt, J. M. S.. Object enhancement and extraction. 1970, New York : Academic Press
[48] Kirsch, R.. Computer Determination of the constituent structure of biological images. Computer and Biomedical Research, 1971, Vol14 (No.3): 315-238
[49] L. S. Davis. A Survey of Edge Detection Techniques. CGIP, 19754: 248-270
[50] R. A. Kirsch. Computer Determination of the Constituent Structure of Biological Images. Computers in Biomedical Research, 1971, 4:315-328
[51] 郭福源．半导体激光束的校正像散变换．激光杂志，1997，Vol．18(No．5)：28～23
[52] 郭福源，翁开华，邱锦辉．半导体激光束的成形变换．光电子·激光，1998，Vol．9(No．1)：43～45
[53] 宋菲君．空间光调制器物理和大屏幕投影电视产业．物理，2000，29(1)
[54] 曹必成．液晶光阀及其应用．光仪技术，1986(4)
[55] W. P. Bleha, L. T. Lipton, E. Wiener-Avnear et al..Opt. Eng.,1978,17(1):371
[56] 陈智勇，赵焕卿，章志鸣．液晶光阀特性的研究，光学学报，1987，7(3)：247～253
[57] 邓志鹏．基于光学小波变换的机器人视觉的研究：[硕士学位论文]．天津：河北工业大学，2000
[58] 王仲民．基于4f光学信息处理的机器人视觉系统设计与研究：[硕士学位论文]．天津：河北工业大学大学，2001：p35
[59] 王文陆．光学子波变换及其在图象处理中的应用：[博士学位论文]．北京：清华大学，1995
[60] 杨福生．小波变换的工程分析与应用．北京：科学出版社，1999
[61] R. Tripathi and K. Singh. Pattern discrimination using a bank of wavelet filters in a joint transform correlator. Opt. Eng., 1998, 37(2): 532～538
[62] Yue Hong, et al. Recognition and Tracking of Large Scan Based on Robotic Vision. Proc. Of the ASME, DSC Vol69-1, pp189-194, 2000
[63] 岳宏等．基于小波变换光学实现机器人视觉传感器．中国国家知识产权局发明专利报，2001，17(30)
[2] 傅京孙，蔡自兴，徐光佑．人工智能及其应用．北京：清华大学出版社，1987
[3] 钟玉琢，乔秉新，李树青．机器人视觉技术．北京：国防工业出社，1994
[4] 郑时雄，刘桂雄，阎华，等．机器人视觉在自动装配线上的应用．光学精密工程，1998，Vol 6(No．1)：90～94
[5] Bonarini, Andrea. Omnidirectioanl vision sensor for fast tracking for mobile robots. IEEE Transactions on Instrumentation and Measurement 49 3 Jun 2000 IEEE p 509-512 0018-9456
[6] 李桂成，卢长刚，王桂琴，等．一种消防机器人的视觉及其信号处理．机器人，1995，Vol．17，No．6．
[7] 王守杰，宗光华，毕树生．微操作机器人显微视觉系统的研究．机器人，1998，Vol．20 (No．6)：138～142
[8] 胡震，袁小海，陈容盛等．用于目标检测和精确定位的水下机器人视觉系统．中国造船，2000，Vol．41(No．2)：89～94
[9] Shimauchi K et al. Developmentof a hockey-robot using real-time vision, Proceedings of the 13th annual conference of RSJ., Kawasaki, 1995: 1137-1138
[10] 赵珠颖，佟国峰．机器人视觉子系统．机器人技术与应用，2000，No．2：26～28
[11] 邬永革，丁震，黄炯，等．一种有效的图象特征抽取和识别方法．计算机研究与发展，1997，Vol．34(No．1)：38～42
[12] 高速信息网络与并行处理研讨会论文集，国家自然科学基金委员会信息科学部，(1994．6)
[13] 清华大学光学仪器教研组．《信息光学基础》．北京：机械：工业出版社，1985
[14] A. Huang. Architectural consideration involved in the design of an optical digital computer..Proc. Of the IEEE, 1984, 72(7): 780-786
[15] A. Greenwood, M. Bartusiak, B. Burke et al. Science at the frontier. Volumel, National Academy Press, 1992
[16] A. W. Lohmann. What classical optics can do for the digital optical computer. Appl. Opt., 1986, 25(10): 1543-1549
[17] 吕乃光．傅立叶光学．北京：机械工业出版社，1988
[18] S. G. Mallat. Zero-Crossings of a wavelet transform. IEEE Trans. Inform. Theory, 1991, IT-37, 1019-1033
[19] O. Rioul and P. Duhamel. Fast algorithms for discrete and Continuous wavelet transforms. 1992, IEEE Trans. Inform. Theory, IT-38(2), 569-586
[20] Lee Jiann Shu, et al. Multiscale Corner Detection by Using Wavelet Transform. IEEE transaction on image processing, 1995, 14(1)
[21] 徐继友，朱耆祥．小波变换在图象边缘提取中的应用．光电工程，1998，Vol．25(No．4)：28～34
[22] 美国科学院，金国藩等．《科学前沿》．1993，第一卷：169～173
[23] 汤焱，莫玉龙．第二代小波变换应用于图象的无损压缩编码．中国图象图形学报，2000，Vol．5(A)(No．8)：699～702
[24] S. J. Schiff. Resolving time-series structure with a controlled Wavelet transform. Opt. Eng. 1992, Vol.31 (No.11): 2492
[25] 翁文国，廖光煊，范维澄．小波变换应用于DPIV图象去噪．中国图象图形学报，2000 Vol．5(A)(No．3)：211～215
[26] M. O. Freeman, A. Fedor, B. Bock et al. Optical Wavelet Processor for producing spatially localized ring-wedged-type information. SPIE Vo1.1772: 241-250(1992)
[27] 靳华，王晓丹，赵荣椿．基于小波变换和ODPSF的纹理图象分割方法．中国图象图形学报，2000 Vol．5(A)(No．10)：857～860
[28] D. Gabor. Theory of communication. J. Inst. Elect. Eng., 1946, Vol.93: 429-457
[29] J. M. Combes, A. Grossman, and Ph. Tchamitchin, eds. Wavelet, Time-Frequency Methods, and Phase Space(Springer-Verlag, New York, 1989): 68-98.
[30] A. Harr. Zur Theorie der orthogonalen Funktionen systeme. Math, Ann. 1910, 69: 331-371
[31] P. Burt and E. Adelson, "The Laplacian pyramid as a compact image code," IEEE Trans. Commun. 31, 532-540(1983)
[32] A. Grossman, J. Morlet. Decomposition of Hardy Function into square Integrable Wavelets of Constant Shape. SLAM. J. Math. Anal, 15(1984)
[33] S. Mallat. Multiresolution approximation and wavelets. Tech. Rep., Dep. Comput. Inform. Sci., Univ. Pennsylvania, Philadephia, PA. Sept(1987).
[34] H. Szu, Yunlong Sheng, and Jing Chen. Wavelet transform as a bank of the matched filters. Appl. Opt., 1992, 31(7): 3267-3277
[35] Guofan Jin, Yingbai Yan, Wenlu Wang et al. Optical Haar Wavelet Transform for Image Features Extraction. Proc. SPIE , 1993, Vol. 2034 Mathematical Imaging, San Diego.
[36] Special issue on Wavelet transform, Opt. Eng. 31, No.9, 1821---1916(1992).
[37] Yao li, and Yan Zhang. Coherent optical processing of Gabor and wavelet expansion of one and two-dimensional signals. Opt. Eng. , 1992, 31(9): 1865-1885
[38] Y. Meyer. In ondelettes et operatears. Paris:Hermann, 1990, pp30-36, 106-110
[39] 马晶，谭立英，冉启文．光学小波滤波理论初探．中国激光，1999，Vol．A26(No．4)：343～346
[40] X. Yang, H. Szu, Y. Sheng, and H. J. Caulfield, Opt. Eng., 31(9), 1846-1851(1992)
[41] T. Burns, K. Fielding, S. Rogers et al. Optical Haar wavelet transform. Opt. Eng., 1992,31(9): 1852-1858.
[42] 王文陆，严瑛白，金国藩，等．光学子波变换(2)实验结构与技术．光电子·激光，1994Vol．5(No．5)：269～272
[43] 康辉，许剑，战元龄，等．利用光折变全息术实现的光学小波变换．中国激光，1996，Vol．A23(No．8)：741～744
[44] 倪明，蒋志平，陆启生．一种光学小波变换的改进．中国激光，1997Vol．A24(No．3)：231～236
[45] 段发阶．计算机视觉检测基础理论及应用技术研究：[博士学位论文]．天津：天津大学，1994
[46] Roberts, L. G.(1965), Machine perception of 3D solids, J.T. Tippett et al eds., Optical and Electrooptical Information Processing, USA: MIT Press: 159-197
[47] Prewitt, J. M. S.. Object enhancement and extraction. 1970, New York : Academic Press
[48] Kirsch, R.. Computer Determination of the constituent structure of biological images. Computer and Biomedical Research, 1971, Vol14 (No.3): 315-238
[49] L. S. Davis. A Survey of Edge Detection Techniques. CGIP, 19754: 248-270
[50] R. A. Kirsch. Computer Determination of the Constituent Structure of Biological Images. Computers in Biomedical Research, 1971, 4:315-328
[51] 郭福源．半导体激光束的校正像散变换．激光杂志，1997，Vol．18(No．5)：28～23
[52] 郭福源，翁开华，邱锦辉．半导体激光束的成形变换．光电子·激光，1998，Vol．9(No．1)：43～45
[53] 宋菲君．空间光调制器物理和大屏幕投影电视产业．物理，2000，29(1)
[54] 曹必成．液晶光阀及其应用．光仪技术，1986(4)
[55] W. P. Bleha, L. T. Lipton, E. Wiener-Avnear et al..Opt. Eng.,1978,17(1):371
[56] 陈智勇，赵焕卿，章志鸣．液晶光阀特性的研究，光学学报，1987，7(3)：247～253
[57] 邓志鹏．基于光学小波变换的机器人视觉的研究：[硕士学位论文]．天津：河北工业大学，2000
[58] 王仲民．基于4f光学信息处理的机器人视觉系统设计与研究：[硕士学位论文]．天津：河北工业大学大学，2001：p35
[59] 王文陆．光学子波变换及其在图象处理中的应用：[博士学位论文]．北京：清华大学，1995
[60] 杨福生．小波变换的工程分析与应用．北京：科学出版社，1999
[61] R. Tripathi and K. Singh. Pattern discrimination using a bank of wavelet filters in a joint transform correlator. Opt. Eng., 1998, 37(2): 532～538
[62] Yue Hong, et al. Recognition and Tracking of Large Scan Based on Robotic Vision. Proc. Of the ASME, DSC Vol69-1, pp189-194, 2000
[63] 岳宏等．基于小波变换光学实现机器人视觉传感器．中国国家知识产权局发明专利报，2001，17(30)
© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号 地址：北京市海淀区学院路29号 邮编：100083 电话：办公室：(+86 10)66554848；文献借阅、咨询服务、科技查新：66554700 |