基于DSP的纸页缺陷检测的关键技术研究
|
摘要
纸页缺陷检测是造纸企业提高自身产品质量,进而增强企业整体竞争力的重要手段。应用计算机视觉和图像处理技术实现对纸页缺陷的在线检测,需要对采集的大量图像数据进行快速实时处理,以便系统能够在实际生产容许的时间范围内完成纸页表面质量信息的精确提取。实时数字图像处理也就成为了实现监测系统实时性要求的关键环节。
近年来,DSP技术的发展不断将数字信号处理领域的理论研究成果应用到实际系统中,并且推动了新的理论和应用领域的发展,对图像处理等领域的技术发展也起到了十分重要的作用。基于DSP的图像处理系统也被广泛的应用于各种领域。
本文通过对现有的纸页检测系统的结构和检测过程的分析,找出影响检测系统实时性的瓶颈,并由此分析提出了基于PC机、图像采集卡和DSP数字图像处理器的纸页缺陷监测系统的设计方案。新的系统的设计从软、硬件两个方面进行。硬件方面,在原有系统中增加了基于TI公司的TMS320C64×DSP的DM642超高速图像处理平台,由DSP完成对影响系统实时性的关键部分的处理;软件方面,提出主从式程序结构和基于图像灰度特征的缺陷图像识别方法,编写了实现缺陷识别的程序,并通过C代码和DSP平台的优化,包括存储器、Cache、EDMA的优化,消除存储器之间的相关性,从原理上减少缓存访问次数,优化循环的软件流水,以此加快程序运行速度,同时通过把Cache L2缓存设置为Cache与SRAM的混合模式实现Cache优化,实现了C代码到DSP平台的移植,同时充分利用TI提供的图像处理库(img62x)和在线仿真技术和软件优化方法对DSP端程序进行了仿真和优化,进而提高软件的运行效率。
本课题研究最终初步完成了对监测系统整体平台的搭建。通过测试,该系统在识别率和实时性方面较原有系统有较大提高,实时处理方面基本满足实际生产的需要。
The surface quality inspection for paper is a most important way for paper corporations to improve their products' quality and thereby enhance their competition capability. The on-line inspection for the surface defect of paper strips using techniques of computer-vision and image-process need to do fleet operation for a great deal of image data being collected. Just with this capability, the system is able to finish collecting quality information of paper strip surface within an accepted time range. Digital image process has become the key point of realization of system real-timeness.
In recent years, the development of DSP technology has put the theory achievement of digital signal processing field into practical application systems and pushed the progress of new theory and application. At the same time, it plays an important role in the field of image processing. The DSP-based image processing system can be applicable in many kinds of practical fields.
The dissertation is based on our lab's research production in the field of the surface defect inspection system of paper. During the analysis of the structure of traditional image processing system and inspection process, the dissertation find the choke point of the system real-timeness, and bring forward the design of surface defect inspection system based on PC host, image-collected card and DSP. The design of new system is carried out both in hardware and in software. In hardware, the traditional system is improved by adding DAM642 whose chip is TMS320C642 of TI. This image process equipment is used to do the part which impacts system real-timeness seriously. In software, principal-and-subordinate software structure is brought. The code running both in DSP and in host is written to do the Cache and EDMA defect inspection. During programming, several new techniques such as on-line emulation and optimization are used to increase the efficiency of system. The final aim of this research is to accomplish on-line surface inspection system for paper strips which is able to fill the supply of industrial manufacture.
The research finally accomplishes primary design of on-line surface inspection system for paper strips. After several tests, the real-timeness and precision of the system has improved obviously, and can satisfy the demand of real production.
At the end of the dissertation, amelioration and expectation for inspection system development based on DSP are mentioned.
近年来,DSP技术的发展不断将数字信号处理领域的理论研究成果应用到实际系统中,并且推动了新的理论和应用领域的发展,对图像处理等领域的技术发展也起到了十分重要的作用。基于DSP的图像处理系统也被广泛的应用于各种领域。
本文通过对现有的纸页检测系统的结构和检测过程的分析,找出影响检测系统实时性的瓶颈,并由此分析提出了基于PC机、图像采集卡和DSP数字图像处理器的纸页缺陷监测系统的设计方案。新的系统的设计从软、硬件两个方面进行。硬件方面,在原有系统中增加了基于TI公司的TMS320C64×DSP的DM642超高速图像处理平台,由DSP完成对影响系统实时性的关键部分的处理;软件方面,提出主从式程序结构和基于图像灰度特征的缺陷图像识别方法,编写了实现缺陷识别的程序,并通过C代码和DSP平台的优化,包括存储器、Cache、EDMA的优化,消除存储器之间的相关性,从原理上减少缓存访问次数,优化循环的软件流水,以此加快程序运行速度,同时通过把Cache L2缓存设置为Cache与SRAM的混合模式实现Cache优化,实现了C代码到DSP平台的移植,同时充分利用TI提供的图像处理库(img62x)和在线仿真技术和软件优化方法对DSP端程序进行了仿真和优化,进而提高软件的运行效率。
本课题研究最终初步完成了对监测系统整体平台的搭建。通过测试,该系统在识别率和实时性方面较原有系统有较大提高,实时处理方面基本满足实际生产的需要。
The surface quality inspection for paper is a most important way for paper corporations to improve their products' quality and thereby enhance their competition capability. The on-line inspection for the surface defect of paper strips using techniques of computer-vision and image-process need to do fleet operation for a great deal of image data being collected. Just with this capability, the system is able to finish collecting quality information of paper strip surface within an accepted time range. Digital image process has become the key point of realization of system real-timeness.
In recent years, the development of DSP technology has put the theory achievement of digital signal processing field into practical application systems and pushed the progress of new theory and application. At the same time, it plays an important role in the field of image processing. The DSP-based image processing system can be applicable in many kinds of practical fields.
The dissertation is based on our lab's research production in the field of the surface defect inspection system of paper. During the analysis of the structure of traditional image processing system and inspection process, the dissertation find the choke point of the system real-timeness, and bring forward the design of surface defect inspection system based on PC host, image-collected card and DSP. The design of new system is carried out both in hardware and in software. In hardware, the traditional system is improved by adding DAM642 whose chip is TMS320C642 of TI. This image process equipment is used to do the part which impacts system real-timeness seriously. In software, principal-and-subordinate software structure is brought. The code running both in DSP and in host is written to do the Cache and EDMA defect inspection. During programming, several new techniques such as on-line emulation and optimization are used to increase the efficiency of system. The final aim of this research is to accomplish on-line surface inspection system for paper strips which is able to fill the supply of industrial manufacture.
The research finally accomplishes primary design of on-line surface inspection system for paper strips. After several tests, the real-timeness and precision of the system has improved obviously, and can satisfy the demand of real production.
At the end of the dissertation, amelioration and expectation for inspection system development based on DSP are mentioned.
引文
[1] J.卫凯西主编, 叶惠莲,李丰年等译. 制浆造纸化学工艺学, 轻工业出版社, 1988.
[2] 曹邦威译. 制浆造纸工程大全(第 2 版). 中国轻工业出版社, 2001.
[3] 吕岑, 孙瑜. 基于统计处理的纸病检测研究. 中国造纸学报, 2003, 18(2): 151-155.
[4] D.Brzakoic and N. Vujovic. Designing a defect classification system: a case study. Pattern Recognition , 1996, 29(8): 1401-1419.
[5] 关健华. 全幅纸病检测技术及在造纸中的应用. 中国造纸, 2000 (G): 32-35.
[6] 温喜东. 纸病检测系统在卫生纸生产的应用. 国际造纸, 2003, 22(6): 4-44.
[7] Jukka Iivarinen and Juhani Rauhamaa. Surface inspection of web materials using the self-organizing map. Proc. SPIE, 1998, 96-103.
[8] D. M. Tsai and S. K. Wu. Automated surface inspection using gabor filters. International Journal ofAdvanced Manufacturing Technology, 2000, (16):474-482.
[9] S. G Mallat. A theory for multiresolution signal decomposition: the wavelet representation. IEEE Trans. Pattern Anal. Mach. Inte11, 1989, (11): 674-693.
[10] T. Chang,C. C. J. Kuo. Texture analysis and classification with tree-structured wavelet transform. IEEE Trans. Image Processing, 1993, 2(4):42-44.
[11] M. Unser. Texture classification and segmentation using wavelet frames. IEEE Trans. Image Processing,1995,4(11): 1549-1560.
[12] G Lambert and F. Bock. Wavelet methods for texture defect detection. IEEE International Conference on Image Processing, 1997,3:201-204.
[13] D. M. Tsai and B. Hsiao. Automatic surface inspection using wavelet recon. Pattern Recognition, 2001,34:1285-1305.
[14] 杨军.ULMA NTi 纸病检测系统.中华纸业,2003,24(3):33-36.
[15] 温喜东.在线纸病检测技术的新发展及应用.中国造纸,2002 (2):61-64.
[16] K. Heikkinen and P. Vuorimaa. Computation of two texture featurehardware. Pro fc. 10th 1nt. Conf. Image Ana 上 ysis and Processing, l999, 125-129.
[17] 雷志勇, 刘群华, 姜寿山, 黄万景, 线阵 CCD 图像处理算法研究[J], 光学技术, 2002.9, 28 (5): 475-478.
[18] 王念旭. DSP 基础与应用系统设计汇[M], 北京航空航天大学出版社, 2000.
[19] 张雄伟, 曹铁勇. DSP 芯片的原理与开发应用[M], 北京:电子工业出版社, 2000 年 9 月第二版,1-25.
[20] 赵威. TMS320C6200 系列 DSP 芯片的应用与开发[M], 北京: 人民邮电出版社, 2002 年 5 月第一版, 1-5.
[21] 沈庭芝, 方子文. 数字图像处理及模式识别[M], 北京: 北京理工大学出版社, 1998 年, 5-15.
[22] 张毓晋. 图像工程[M], 北京:清华大学出版社, 2000 年.
[23] Andrew Bateman, lains Paterson-Stephens 著, 陈健, 陈伟, 汪书宁译. DSP 算法、应用与设计[M], 北京:机械工业出版社, 中信出版社, 2004 年 10 月第一版, 28-29.
[24] 北京闻亭科技发展有限公司. DAM6416P 图像处理平台用户使用手册, 2003年 9 月, 7-8.
[25] Texas Instruments Co. TMS320C6000 Technical BrieP,February 1999.
[26] Texas Instruments Co.Code Composer Studio User's Guide,February 1998.
[27] Texas Instruments Co.TMS320C64x Technical Overview, January 2001.
[28] 李贵山. PCI 局部总线开发者指南,西安电子科技大学出版社[M], 1997 年 1 月
[29] 尹勇, 欧光军, 关荣锋. DSP 集成开发环境 CCS 开发指南[M], 北京航空航天大学出版社, 2003 年 11 月第一版.
[30] Texas Instruments Co.Code Composer Studio Started Guide,November 2001.
[31] Texas Instruments Co.TMS320 DSPBIOS User's Guide, April 2003.
[32] Texas Instruments Co.TMS320C6000 DSP BIOS Application Prog Interfa (API) Reference Guide, April 2003.
[33] 北京闻亭科技发展有限责任公司. TDS560USB 实时仿真系统安装指南, 2005年 4 月.
[34] 北京闻亭科技发展有限责任公司. DAM6416 图像处理平台软件开发指南, 2003 年 9 月.
[35] Texas Instruments . TMS320C64x DSP Programmers' Reference, April 2002.
[36] 龙腾科技. Visual C++ 6. 循序渐进教程[M], 北京: 红旗出版社,北京希望出版社, 2005 年 2 月第一版.
[37] 夏云龙.最新 Visual C++使用手册[M], 北京:电子工业出版社, 2005 年 1 月
[38] 何斌, 马天予, 王运坚, 朱红莲.Visual C++数字图像处理(第二版)[M], 北京:人民邮电出版社, 2002 年 12 月.
[39] 郎锐.数字图像处理学 VC++实现, 北京希望电子出版社, 2003 年 1 月第一版.
[40] Texas Instruments Co. TMS320C6000 Programmer's Guide, August 2002.
[41] 刘栋. 基于机器视觉的板带钢图像实时采集与处理系统研究[D], 沈阳: 东北大学, 2004 年 12 月.
[42] 田捷, 沙飞, 张新生.实用图像分析与处理技术[M], 电子工业出版社, 1995.
[43] Prithwijit Guha.Automated Visual Inspection of Steel Surface,Texture Seg and Development of A Perceptual Similarity Measure:[A Thesis Submitted for the Degree of Master of Technology],KANPUR:INDIAN INSTITUTE OF TE,2001.
[44] Texas Instruments Co.TMS320C6000 Compiler User's Gui,October 2002
[45] 衰金丽, 吉海彦, 郭志涛.基于计算机视觉的苹果表而缺陷识别方法. 农业网络信息, 2005, (6):10-17.
[46] 蔡叶篙. 铝塑泡装药.b.包装检测系统. 中国包装工业, 2003.5:3839.
[47] 安永泉, 郭圣权. 曲线拟合在图像恢复中的应用. 华北土学院测试技术学报. 1999, 13(4):231-234.
[48] 陈良泽. 用矩阵运算曲线拟合的最小二乘法. 传感器技术, 2001, 20(2):3034
[49] 白刚.纱线的嵌入及织物的旋转.大连轻土业学院,硕士学位论文,2002:5356.
[50] 纪钢, 肖朝晖. 国产.bI.包装外观图像信息数字化及数据库技术的应用.包装工程, 2005, 26(4):104-105.
[51] 王德忠, 王涛. 计算机技术在包装领域中的应用. 包装工程, 1998, 19(2):1-3.
[52] 余洪山, 王耀南. 一种改进型边缘检测算法.计算机工程与应用, 2004, 20:2729.
[53] 王植, 贺赛先. 一种基于 Canny 理论的自适应边缘检测方法. 中国图像图形学报, 2004, 9(8): 957-962.
[54] 杨枝灵, 王开等. VC++数字图像获取实践应用. 人民邮电出版社, 2003: 2600.
[2] 曹邦威译. 制浆造纸工程大全(第 2 版). 中国轻工业出版社, 2001.
[3] 吕岑, 孙瑜. 基于统计处理的纸病检测研究. 中国造纸学报, 2003, 18(2): 151-155.
[4] D.Brzakoic and N. Vujovic. Designing a defect classification system: a case study. Pattern Recognition , 1996, 29(8): 1401-1419.
[5] 关健华. 全幅纸病检测技术及在造纸中的应用. 中国造纸, 2000 (G): 32-35.
[6] 温喜东. 纸病检测系统在卫生纸生产的应用. 国际造纸, 2003, 22(6): 4-44.
[7] Jukka Iivarinen and Juhani Rauhamaa. Surface inspection of web materials using the self-organizing map. Proc. SPIE, 1998, 96-103.
[8] D. M. Tsai and S. K. Wu. Automated surface inspection using gabor filters. International Journal ofAdvanced Manufacturing Technology, 2000, (16):474-482.
[9] S. G Mallat. A theory for multiresolution signal decomposition: the wavelet representation. IEEE Trans. Pattern Anal. Mach. Inte11, 1989, (11): 674-693.
[10] T. Chang,C. C. J. Kuo. Texture analysis and classification with tree-structured wavelet transform. IEEE Trans. Image Processing, 1993, 2(4):42-44.
[11] M. Unser. Texture classification and segmentation using wavelet frames. IEEE Trans. Image Processing,1995,4(11): 1549-1560.
[12] G Lambert and F. Bock. Wavelet methods for texture defect detection. IEEE International Conference on Image Processing, 1997,3:201-204.
[13] D. M. Tsai and B. Hsiao. Automatic surface inspection using wavelet recon. Pattern Recognition, 2001,34:1285-1305.
[14] 杨军.ULMA NTi 纸病检测系统.中华纸业,2003,24(3):33-36.
[15] 温喜东.在线纸病检测技术的新发展及应用.中国造纸,2002 (2):61-64.
[16] K. Heikkinen and P. Vuorimaa. Computation of two texture featurehardware. Pro fc. 10th 1nt. Conf. Image Ana 上 ysis and Processing, l999, 125-129.
[17] 雷志勇, 刘群华, 姜寿山, 黄万景, 线阵 CCD 图像处理算法研究[J], 光学技术, 2002.9, 28 (5): 475-478.
[18] 王念旭. DSP 基础与应用系统设计汇[M], 北京航空航天大学出版社, 2000.
[19] 张雄伟, 曹铁勇. DSP 芯片的原理与开发应用[M], 北京:电子工业出版社, 2000 年 9 月第二版,1-25.
[20] 赵威. TMS320C6200 系列 DSP 芯片的应用与开发[M], 北京: 人民邮电出版社, 2002 年 5 月第一版, 1-5.
[21] 沈庭芝, 方子文. 数字图像处理及模式识别[M], 北京: 北京理工大学出版社, 1998 年, 5-15.
[22] 张毓晋. 图像工程[M], 北京:清华大学出版社, 2000 年.
[23] Andrew Bateman, lains Paterson-Stephens 著, 陈健, 陈伟, 汪书宁译. DSP 算法、应用与设计[M], 北京:机械工业出版社, 中信出版社, 2004 年 10 月第一版, 28-29.
[24] 北京闻亭科技发展有限公司. DAM6416P 图像处理平台用户使用手册, 2003年 9 月, 7-8.
[25] Texas Instruments Co. TMS320C6000 Technical BrieP,February 1999.
[26] Texas Instruments Co.Code Composer Studio User's Guide,February 1998.
[27] Texas Instruments Co.TMS320C64x Technical Overview, January 2001.
[28] 李贵山. PCI 局部总线开发者指南,西安电子科技大学出版社[M], 1997 年 1 月
[29] 尹勇, 欧光军, 关荣锋. DSP 集成开发环境 CCS 开发指南[M], 北京航空航天大学出版社, 2003 年 11 月第一版.
[30] Texas Instruments Co.Code Composer Studio Started Guide,November 2001.
[31] Texas Instruments Co.TMS320 DSPBIOS User's Guide, April 2003.
[32] Texas Instruments Co.TMS320C6000 DSP BIOS Application Prog Interfa (API) Reference Guide, April 2003.
[33] 北京闻亭科技发展有限责任公司. TDS560USB 实时仿真系统安装指南, 2005年 4 月.
[34] 北京闻亭科技发展有限责任公司. DAM6416 图像处理平台软件开发指南, 2003 年 9 月.
[35] Texas Instruments . TMS320C64x DSP Programmers' Reference, April 2002.
[36] 龙腾科技. Visual C++ 6. 循序渐进教程[M], 北京: 红旗出版社,北京希望出版社, 2005 年 2 月第一版.
[37] 夏云龙.最新 Visual C++使用手册[M], 北京:电子工业出版社, 2005 年 1 月
[38] 何斌, 马天予, 王运坚, 朱红莲.Visual C++数字图像处理(第二版)[M], 北京:人民邮电出版社, 2002 年 12 月.
[39] 郎锐.数字图像处理学 VC++实现, 北京希望电子出版社, 2003 年 1 月第一版.
[40] Texas Instruments Co. TMS320C6000 Programmer's Guide, August 2002.
[41] 刘栋. 基于机器视觉的板带钢图像实时采集与处理系统研究[D], 沈阳: 东北大学, 2004 年 12 月.
[42] 田捷, 沙飞, 张新生.实用图像分析与处理技术[M], 电子工业出版社, 1995.
[43] Prithwijit Guha.Automated Visual Inspection of Steel Surface,Texture Seg and Development of A Perceptual Similarity Measure:[A Thesis Submitted for the Degree of Master of Technology],KANPUR:INDIAN INSTITUTE OF TE,2001.
[44] Texas Instruments Co.TMS320C6000 Compiler User's Gui,October 2002
[45] 衰金丽, 吉海彦, 郭志涛.基于计算机视觉的苹果表而缺陷识别方法. 农业网络信息, 2005, (6):10-17.
[46] 蔡叶篙. 铝塑泡装药.b.包装检测系统. 中国包装工业, 2003.5:3839.
[47] 安永泉, 郭圣权. 曲线拟合在图像恢复中的应用. 华北土学院测试技术学报. 1999, 13(4):231-234.
[48] 陈良泽. 用矩阵运算曲线拟合的最小二乘法. 传感器技术, 2001, 20(2):3034
[49] 白刚.纱线的嵌入及织物的旋转.大连轻土业学院,硕士学位论文,2002:5356.
[50] 纪钢, 肖朝晖. 国产.bI.包装外观图像信息数字化及数据库技术的应用.包装工程, 2005, 26(4):104-105.
[51] 王德忠, 王涛. 计算机技术在包装领域中的应用. 包装工程, 1998, 19(2):1-3.
[52] 余洪山, 王耀南. 一种改进型边缘检测算法.计算机工程与应用, 2004, 20:2729.
[53] 王植, 贺赛先. 一种基于 Canny 理论的自适应边缘检测方法. 中国图像图形学报, 2004, 9(8): 957-962.
[54] 杨枝灵, 王开等. VC++数字图像获取实践应用. 人民邮电出版社, 2003: 2600.