烟草异物在线检测与实时剔除系统的研究与应用
|
摘要
烟草行业作为一个特殊的产业，为世界经济的发展做出了巨大贡献，成为了各国经济财政收入的重要来源。然而，随着人们生活水平的不断提高，人们对自身健康状况的关注日益加强，吸烟有害健康这一思想已深入人心，如何降低烟草对人的危害程度，提高香烟的品质与质量，已经成为烟草行业长足发展的首选途径。在世界经济一体化进程逐步推进的今天，对利润丰厚的烟草行业，更成为世界各国的各大烟草公司追逐的焦点。要想在日益激烈的市场竞争中获取胜利，必须提高公司产品的竞争力。提高烟草产品的质量，减少对人的伤害，最重要的环节就是在烟草生产过程中的异物剔除问题。
在传统的生产工艺中，异物剔除这一环节是通过人们手工来完成，这种方式效率低下，而且可靠性极低。为了提高企业竞争力，不少大型烟草企业都想使用高自动化的异物剔除设备来完成异物的自动剔除。然而，目前这种大型的高自动化的异物剔除设备在我国国内还没有研制与开发的成功经验，更不可能有满意的产品以供选择，获取的唯一途径便是通过进口。但是引进这种高技术含量的设备除了价格昂贵，技术支持困难等因数外，而且更重要的是受知识产权的限制，根本不可能引进到与国外同期水平的该类产品，因而，也不可能与国外烟草企业进行平等的竞争。因此，研制出具有我国自主知识产权的异物剔除设备具有十分深远的意义。
由重庆大学和昆明船舶集团公司联合研制的烟草异物在线检测与实时剔除系统就是在国家计委专项基金的支持下，结合了国内外同行研制与开发的经验，自行独立研制的具有自主知识产权的异物剔除设备。该系统的工作原理为：由高性能彩色线阵CCD摄像机获取高速运动的烟草物流图像，通过实时测控软件的处理，完成异物识别与异物相对位置信息的获取，最后将异物位置信息发送给执行机构，由执行机构来完成异物剔除的最终目的。该系统的关键是要解决大数据量处理的实时性和对多种异物不同批次烟叶的准确判别这两个问题。在通用操作系统Windows 2000中解决工业用高实时性问题不是一件容易的事情，笔者在结合相应软件开发包的基础上，实现对物理内存的直接访问和操作，完成系统的实时性要求。针对不同的烟草和大量不可预知的异物，设计了浮动阈值和自适应机敏阈值判别方法，同时结合各种不同的色彩空间进行处理，完成对多达18种异物的精确识别和剔除。
本系统原理样机在缺少风压护送系统的作用下实现了光、机、电、气以及计算机控制的一体化联调，实现了自动送料、物料均散、高速输送（5m/s）、动态图
像采集、高速数据传输、异物识别、异物剔除等功能。原理样机在传送带速度为5m/s左右时，用烟草中可能混合的各种异物进行测试，结果表明：基本上能对各种异物进行识别和剔除。原理样机在异物单独测试的情形下，对30mm×30mm异物的综合识别率为78.7℅，剔除率为62.4℅，验证了本系统设计方案的可行性。
在大异物30mm×30mm处理与识别时，本系统能够达到较高的技术水平，但在小异物的识别与处理上，与实际产品的指标还有一定距离。经过后期的数据分析和讨论，找到了提高识别精度的办法，作为后期研制的主要切入点。首先，提高线阵CCD的工作性能，采用棱镜分光实现R、G、B三色同时感光，彻底消除三色错位问题；其次，采用更优质的采集卡，解决采集卡的最小采集单元(目前只能实现最小采集16线)的问题。最后，综合更简洁有效处理与识别算法以及提高光源稳定性等。
The tobacco, as a special industry, has contributed to the economic development in the world greatly, and become the important source of the finance income of all the countries. However, since the people's life level increases continuously, people pay more attention to health condition of their own. The idea that smoking is harmful to health has gone deeply into the heart gradually. Then, how to reduce the harmfulness of tobacco to people, how to increase the quality of the cigarette, all above has become the first way to develop tobacco industry. Because tobacco industry can make a great fortunate, it has already become the focus that pursued by international tobacco company. In order to win the competition, we should enhance the competitive capability of the products, improve the quantity of the tobacco product and reduce disservice to people. Concretely speaking, eliminating the foreign body in the tobacco production line is the most important means.
Traditionally, eliminating the foreign body is completed manually. Certainly the efficiency and dependence of this way is rather lower. Now a few large tobacco enterprises want to use high automation equipment to complete this work automatically so as to enhance competition capability of enterprise. However, this kind of automation equipment has not manufactured in home with successful experience, the only way is to import. But the advanced equipment with high technique is very expensive and its technique support is not available, most important is that it is impossible to introduce product in same period because of the intelligent property right. Therefore, it has the very profound meaning to tap and manufacture this kind of high automation equipment with the intelligent property right.
Foreign body in tobacco online detecting and real-time eliminating system, supported under the country plan fund, is taped and manufactured by Chongqing university and Kunming ship group LTD. The work principle of system is following: Using high speed color line CCD to obtain the image of tobacco flow, through special software's processing, to recognize and locate the foreign body in tobacco, finally send out the foreign body position information to nozzles, which eliminate the foreign body in the end. The key point of the system is two sides, one is real-time processing, and the other is that exactly recognizing various foreign bodies in tobacco. Real-time processing is not an easy case based on Microsoft windows 2000, the writer combined
the basic of the relevant SDK, completed not only the kernel programming, but also physical memory directly access. Aimed at the different tobacco with the random chroma value, self-adapting threshold algorithm is designed. Combined with various color space mode to process at the same time, completed precise recognition and location more than 18 kinds of foreign bodies.
The prototype has unified the optical, mechanical, electronical, aerodynamics and computer control under the lack of wind press system, and be able to send material automatically, make the material spread, transport the material in a high speed (5m/s), grab image dynamically, transmit data with super-speed, recognize the foreign body, eliminate the foreign body. The successful process is validated the feasibility of whole system. In such condition, to single 30mm × 30mm foreign body, the recognition rate is 78.7℅ and the eliminating rate is 62.4℅. It is illustrated the feasibility of the design principle.
To large foreign body such as 30mm×30 mm, the performance of system is rather good, but to small foreign body, the performance is not satisfaction. We analyze the experimental data and find out the way of improvement. In addition to improve wind press system, still do effort in other aspects. First, high performance line CCD, adopted prisms divide light into R, G, B three beams without three color-offset problem. Secondly, adopted the sophisticated grab card to solve minimal unit (the current card can realize the least 16 lines). Finally, synthesize more succin
在传统的生产工艺中，异物剔除这一环节是通过人们手工来完成，这种方式效率低下，而且可靠性极低。为了提高企业竞争力，不少大型烟草企业都想使用高自动化的异物剔除设备来完成异物的自动剔除。然而，目前这种大型的高自动化的异物剔除设备在我国国内还没有研制与开发的成功经验，更不可能有满意的产品以供选择，获取的唯一途径便是通过进口。但是引进这种高技术含量的设备除了价格昂贵，技术支持困难等因数外，而且更重要的是受知识产权的限制，根本不可能引进到与国外同期水平的该类产品，因而，也不可能与国外烟草企业进行平等的竞争。因此，研制出具有我国自主知识产权的异物剔除设备具有十分深远的意义。
由重庆大学和昆明船舶集团公司联合研制的烟草异物在线检测与实时剔除系统就是在国家计委专项基金的支持下，结合了国内外同行研制与开发的经验，自行独立研制的具有自主知识产权的异物剔除设备。该系统的工作原理为：由高性能彩色线阵CCD摄像机获取高速运动的烟草物流图像，通过实时测控软件的处理，完成异物识别与异物相对位置信息的获取，最后将异物位置信息发送给执行机构，由执行机构来完成异物剔除的最终目的。该系统的关键是要解决大数据量处理的实时性和对多种异物不同批次烟叶的准确判别这两个问题。在通用操作系统Windows 2000中解决工业用高实时性问题不是一件容易的事情，笔者在结合相应软件开发包的基础上，实现对物理内存的直接访问和操作，完成系统的实时性要求。针对不同的烟草和大量不可预知的异物，设计了浮动阈值和自适应机敏阈值判别方法，同时结合各种不同的色彩空间进行处理，完成对多达18种异物的精确识别和剔除。
本系统原理样机在缺少风压护送系统的作用下实现了光、机、电、气以及计算机控制的一体化联调，实现了自动送料、物料均散、高速输送（5m/s）、动态图
像采集、高速数据传输、异物识别、异物剔除等功能。原理样机在传送带速度为5m/s左右时，用烟草中可能混合的各种异物进行测试，结果表明：基本上能对各种异物进行识别和剔除。原理样机在异物单独测试的情形下，对30mm×30mm异物的综合识别率为78.7℅，剔除率为62.4℅，验证了本系统设计方案的可行性。
在大异物30mm×30mm处理与识别时，本系统能够达到较高的技术水平，但在小异物的识别与处理上，与实际产品的指标还有一定距离。经过后期的数据分析和讨论，找到了提高识别精度的办法，作为后期研制的主要切入点。首先，提高线阵CCD的工作性能，采用棱镜分光实现R、G、B三色同时感光，彻底消除三色错位问题；其次，采用更优质的采集卡，解决采集卡的最小采集单元(目前只能实现最小采集16线)的问题。最后，综合更简洁有效处理与识别算法以及提高光源稳定性等。
The tobacco, as a special industry, has contributed to the economic development in the world greatly, and become the important source of the finance income of all the countries. However, since the people's life level increases continuously, people pay more attention to health condition of their own. The idea that smoking is harmful to health has gone deeply into the heart gradually. Then, how to reduce the harmfulness of tobacco to people, how to increase the quality of the cigarette, all above has become the first way to develop tobacco industry. Because tobacco industry can make a great fortunate, it has already become the focus that pursued by international tobacco company. In order to win the competition, we should enhance the competitive capability of the products, improve the quantity of the tobacco product and reduce disservice to people. Concretely speaking, eliminating the foreign body in the tobacco production line is the most important means.
Traditionally, eliminating the foreign body is completed manually. Certainly the efficiency and dependence of this way is rather lower. Now a few large tobacco enterprises want to use high automation equipment to complete this work automatically so as to enhance competition capability of enterprise. However, this kind of automation equipment has not manufactured in home with successful experience, the only way is to import. But the advanced equipment with high technique is very expensive and its technique support is not available, most important is that it is impossible to introduce product in same period because of the intelligent property right. Therefore, it has the very profound meaning to tap and manufacture this kind of high automation equipment with the intelligent property right.
Foreign body in tobacco online detecting and real-time eliminating system, supported under the country plan fund, is taped and manufactured by Chongqing university and Kunming ship group LTD. The work principle of system is following: Using high speed color line CCD to obtain the image of tobacco flow, through special software's processing, to recognize and locate the foreign body in tobacco, finally send out the foreign body position information to nozzles, which eliminate the foreign body in the end. The key point of the system is two sides, one is real-time processing, and the other is that exactly recognizing various foreign bodies in tobacco. Real-time processing is not an easy case based on Microsoft windows 2000, the writer combined
the basic of the relevant SDK, completed not only the kernel programming, but also physical memory directly access. Aimed at the different tobacco with the random chroma value, self-adapting threshold algorithm is designed. Combined with various color space mode to process at the same time, completed precise recognition and location more than 18 kinds of foreign bodies.
The prototype has unified the optical, mechanical, electronical, aerodynamics and computer control under the lack of wind press system, and be able to send material automatically, make the material spread, transport the material in a high speed (5m/s), grab image dynamically, transmit data with super-speed, recognize the foreign body, eliminate the foreign body. The successful process is validated the feasibility of whole system. In such condition, to single 30mm × 30mm foreign body, the recognition rate is 78.7℅ and the eliminating rate is 62.4℅. It is illustrated the feasibility of the design principle.
To large foreign body such as 30mm×30 mm, the performance of system is rather good, but to small foreign body, the performance is not satisfaction. We analyze the experimental data and find out the way of improvement. In addition to improve wind press system, still do effort in other aspects. First, high performance line CCD, adopted prisms divide light into R, G, B three beams without three color-offset problem. Secondly, adopted the sophisticated grab card to solve minimal unit (the current card can realize the least 16 lines). Finally, synthesize more succin
引文
[1] 聂和平.加快推进烟草市场化进程 迎接经济全球化的挑战.中国烟草科学.2001.4:12-14
[2] 任炜. 论烟草农业科学技术可持续发展. 云南烟草学会1998年年会文件汇编.1998.
[3] 石耀东.我国烟草工业面临的主要问题与出路.中国工业经济.1999.2
[4] 郭文川 王转卫 朱新华.机器视觉技术在果蔬的识别与分级中的应用.粮油加工与食品机械.2002.5
[5] 应义斌 饶秀勤 赵匀.机器视觉技术在农产品品质自动识别中的应用研究进展.农业工程学报.2000.16(3)
[6] Operator's manual of TobaccoScan HS.Hauni Co.LTD.1999
[7] ANMERSON D.Manufacturing with vision[J].Word Tobacco.1999.7
[8] 黄秋红.宽照度范围的CCD在线检测技术的实验研究[学位论文].西安.西安理工大学.2000.3
[9] Feng Yuan. Windows Graphics Programming Win32 GDI and DirectDraw.Beijing.China Machine Press.2002.4
[10] 傅德胜 寿益禾.图形图像处理学.第一版.南京.东南大学出版社.2001.12
[11] 苏光大.微机图像处理系统.第一版.北京.清华大学出版社.2000.7
[12] 张盛彬,等. 彩色线阵CCD用于物体尺寸精密测量的研究. 光电子·激光. 2001,12(11):1159-1162
[13] 周 鸿,等. 基于线阵CCD阵列的在线检测方法 [J] 半导体技术. 2001,26(11): 1-4
[14] 方如明 蔡键荣 许俐.计算机图像处理技术及其在农业工程中的应用.第一版.北京.清华大学出版社.200
[15] 章毓晋.图像工程.第一版.北京.清华大学出版社.1998
[16] 边肇祺.模式识别.第二版.北京.清华大学出版社.1985
[17] ArthurR.Weeks,G.Eric Hague.Color Segmentation in the HIS color space using the K-means algorithm[J]. SPIE 1997(30)
[18] Medford Oregon.97501 USA.SRC VISION Part No.95-169291-001.1997.Tobacco Sorter Korma-sort Controller
[19] Medford Oregon.97501 USA.SRC VISION Part No.95-169845-001.1997.Tobacco Sorter Korma-sort Controller
[20] Medford Oregon.97501 USA.SRC VISION Part No.95-165846-001.1997.Tobacco Sorter Korma-sort Controller
[21] John H.Wilbur Todd Hoffman. Off-Belt Stabilizing system for light weight articles.
USA.6003681.Dec.21, 1999
[22] 金洪震,等. 线阵CCD光谱特性的研究及应用 [J] 浙江师大学报（自然科学版）. 2000,23(3): 53-256
[23] 袁祥辉.固体图像传感器及其应用.重庆.重庆大学出版社.1996
[24] BASLER L301BC installation Guide.Basler AG.2000
[25] BASLER CCT installation Guide.Basler AG.2000
[26] 陈杰 李志敏 钟先信等.烟草在线检测与异物剔除系统.光电工程.2003.3
[27] 王险峰 刘宝宏.Windows环境下的多线程编程原理与应用.第一版.北京.清华大学出版社.2002.6
[28] 曹子元等.Win32下物理内存分配和直接访问的实现.测控技术.2001,20(1):46-47
[29] 郑学坚 周斌.微型计算机原理及应用.第二版.北京.清华大学出版社.1998.1
[30] 陈杰 李志敏 钟先信等.高速物流图像采集与实时异物剔除的原理及实现.光学 精密工程.2002.10（5）:454-458
[31] [美]Jeffrey Richter.王书洪 刘光明.Windows高级编程指南.第三版.北京.清华大学出版社.1999.6
[32] 木森林 高峰霞等.Windows95实用编程与规范.第一版.北京.清华大学出版社.1997.6
[33] Matrox Group.Matrox Image Library User Guide [M].Canda.Matrox Press.2000
[34] RICHTER J.Program Applications for Windows[M].USA.Microsoft Press.1999
[35] Intel.Intel Memory Products[M].USA.Intel Press.1991
[36] 周德兴等.实时系统软件设计方法.计算机自动测量与控制.2000.4:4-5
[37] 李志敏 陈杰等.高速彩色线阵CCD三色错位修正方法的研究.仪器仪表学报（增刊）.2003.4
在校期间发表论文
[01] 陈杰 李志敏 钟先信等.高速物流图像采集与实时异物剔除的原理及实现.光学精密工程，2002，11（5）：454－458
[02] 陈杰 李志敏 钟先信等.烟草在线检测与异物剔除系统.光电工程.已录用2003.3发表
[03] 李志敏 陈杰等. 高速彩色线阵CCD三色错位修正方法的研究.仪器仪表学报，已录用.2003.4（增刊）发表
[04] 李志敏 吴建军 陈杰等.计算机图像处理技术在晶粒度测量中的应用.压电与声光.2002,24(2):168-170
[05] 李志敏 李育才 陈杰等.采用FPS110的指纹识别仪.半导体光电. 2002,23(3):198-200
[06] 陈文涛 钟先信 李志敏 刘军 陈杰.烟草异物在线检测系统.工业仪表与自动化装置.2003，（2）57－59
[07] 陈文涛 钟先信 李志敏 刘军 陈杰.烟草在线异物剔除系统研究.光学精密工程.已录用，2003年未定期发表
[08] 陈文涛 钟先信 李志敏 刘军 陈杰.烟草异物在线检测系统设计.仪器技术与传感器.已录用（Ya-02-108）
[09] 陈文涛 钟先信 李志敏 刘军 陈杰.烟草在线异物剔除系统中的关键技术.测控技术.已录用,2003，22（5）发表
[10] 刘军 钟先信 李志敏 陈文涛 陈杰.在线烟草异物剔除系统.计算机测量与控制，2003.10(2):95-97
[2] 任炜. 论烟草农业科学技术可持续发展. 云南烟草学会1998年年会文件汇编.1998.
[3] 石耀东.我国烟草工业面临的主要问题与出路.中国工业经济.1999.2
[4] 郭文川 王转卫 朱新华.机器视觉技术在果蔬的识别与分级中的应用.粮油加工与食品机械.2002.5
[5] 应义斌 饶秀勤 赵匀.机器视觉技术在农产品品质自动识别中的应用研究进展.农业工程学报.2000.16(3)
[6] Operator's manual of TobaccoScan HS.Hauni Co.LTD.1999
[7] ANMERSON D.Manufacturing with vision[J].Word Tobacco.1999.7
[8] 黄秋红.宽照度范围的CCD在线检测技术的实验研究[学位论文].西安.西安理工大学.2000.3
[9] Feng Yuan. Windows Graphics Programming Win32 GDI and DirectDraw.Beijing.China Machine Press.2002.4
[10] 傅德胜 寿益禾.图形图像处理学.第一版.南京.东南大学出版社.2001.12
[11] 苏光大.微机图像处理系统.第一版.北京.清华大学出版社.2000.7
[12] 张盛彬,等. 彩色线阵CCD用于物体尺寸精密测量的研究. 光电子·激光. 2001,12(11):1159-1162
[13] 周 鸿,等. 基于线阵CCD阵列的在线检测方法 [J] 半导体技术. 2001,26(11): 1-4
[14] 方如明 蔡键荣 许俐.计算机图像处理技术及其在农业工程中的应用.第一版.北京.清华大学出版社.200
[15] 章毓晋.图像工程.第一版.北京.清华大学出版社.1998
[16] 边肇祺.模式识别.第二版.北京.清华大学出版社.1985
[17] ArthurR.Weeks,G.Eric Hague.Color Segmentation in the HIS color space using the K-means algorithm[J]. SPIE 1997(30)
[18] Medford Oregon.97501 USA.SRC VISION Part No.95-169291-001.1997.Tobacco Sorter Korma-sort Controller
[19] Medford Oregon.97501 USA.SRC VISION Part No.95-169845-001.1997.Tobacco Sorter Korma-sort Controller
[20] Medford Oregon.97501 USA.SRC VISION Part No.95-165846-001.1997.Tobacco Sorter Korma-sort Controller
[21] John H.Wilbur Todd Hoffman. Off-Belt Stabilizing system for light weight articles.
USA.6003681.Dec.21, 1999
[22] 金洪震,等. 线阵CCD光谱特性的研究及应用 [J] 浙江师大学报（自然科学版）. 2000,23(3): 53-256
[23] 袁祥辉.固体图像传感器及其应用.重庆.重庆大学出版社.1996
[24] BASLER L301BC installation Guide.Basler AG.2000
[25] BASLER CCT installation Guide.Basler AG.2000
[26] 陈杰 李志敏 钟先信等.烟草在线检测与异物剔除系统.光电工程.2003.3
[27] 王险峰 刘宝宏.Windows环境下的多线程编程原理与应用.第一版.北京.清华大学出版社.2002.6
[28] 曹子元等.Win32下物理内存分配和直接访问的实现.测控技术.2001,20(1):46-47
[29] 郑学坚 周斌.微型计算机原理及应用.第二版.北京.清华大学出版社.1998.1
[30] 陈杰 李志敏 钟先信等.高速物流图像采集与实时异物剔除的原理及实现.光学 精密工程.2002.10（5）:454-458
[31] [美]Jeffrey Richter.王书洪 刘光明.Windows高级编程指南.第三版.北京.清华大学出版社.1999.6
[32] 木森林 高峰霞等.Windows95实用编程与规范.第一版.北京.清华大学出版社.1997.6
[33] Matrox Group.Matrox Image Library User Guide [M].Canda.Matrox Press.2000
[34] RICHTER J.Program Applications for Windows[M].USA.Microsoft Press.1999
[35] Intel.Intel Memory Products[M].USA.Intel Press.1991
[36] 周德兴等.实时系统软件设计方法.计算机自动测量与控制.2000.4:4-5
[37] 李志敏 陈杰等.高速彩色线阵CCD三色错位修正方法的研究.仪器仪表学报（增刊）.2003.4
在校期间发表论文
[01] 陈杰 李志敏 钟先信等.高速物流图像采集与实时异物剔除的原理及实现.光学精密工程，2002，11（5）：454－458
[02] 陈杰 李志敏 钟先信等.烟草在线检测与异物剔除系统.光电工程.已录用2003.3发表
[03] 李志敏 陈杰等. 高速彩色线阵CCD三色错位修正方法的研究.仪器仪表学报，已录用.2003.4（增刊）发表
[04] 李志敏 吴建军 陈杰等.计算机图像处理技术在晶粒度测量中的应用.压电与声光.2002,24(2):168-170
[05] 李志敏 李育才 陈杰等.采用FPS110的指纹识别仪.半导体光电. 2002,23(3):198-200
[06] 陈文涛 钟先信 李志敏 刘军 陈杰.烟草异物在线检测系统.工业仪表与自动化装置.2003，（2）57－59
[07] 陈文涛 钟先信 李志敏 刘军 陈杰.烟草在线异物剔除系统研究.光学精密工程.已录用，2003年未定期发表
[08] 陈文涛 钟先信 李志敏 刘军 陈杰.烟草异物在线检测系统设计.仪器技术与传感器.已录用（Ya-02-108）
[09] 陈文涛 钟先信 李志敏 刘军 陈杰.烟草在线异物剔除系统中的关键技术.测控技术.已录用,2003，22（5）发表
[10] 刘军 钟先信 李志敏 陈文涛 陈杰.在线烟草异物剔除系统.计算机测量与控制，2003.10(2):95-97