残损胶囊图像检测系统的研究与设计
|
摘要
随着现代经济与科学的快速发展,人们生活水平不断的提高。与此同时,人们对与自身健康息息相关的药品质量有了越来越高的质检要求。在日常生活中我们常见的胶囊药品在购买时可以凭借肉眼获取药品的剂量与外形轮廓、色泽等物理量来判别胶囊药品质量的好坏。但是在许多制药厂药品的批量生产过程中单纯的靠员工的眼睛来逐个判断药品的外形轮廓与色泽,从而不能满足现代工业生产的要求,这样不仅增加了生产与管理成本同时由于人眼的不稳定性与易疲劳等缺点无法保障高的检测合格率。基于这样的背景,为了完善制药厂现有设备的检测配套功能,本文在系统描述了现有设备的运行方式及要求的基础上,结合数字图像处理技术,对胶囊分拣系统的核心部分—硬件设计、破损胶囊检测算法、系统软件设计进行了研究,并进行了算法的仿真验证。
在硬件方面,根据胶囊检测设备的实时性要求,本文提出了一套由DSP+FPGA组成的胶囊图像检测系统解决方案,并重点介绍了CCD摄像头的构成及工作原理、视频信号转换、FPGA配置方式及其在系统设计中的功能、DSP的外部硬件连接等相关内容。
在胶囊检测算法方面,介绍了基本的数字图像处理理论知识及常用的研究方法。本文引入了灰色系统理论作为数字图像处理的手段,利用AGO数据生成和GM(1,1)模型设计了一套数字图像滤波算法。在数字图像处理理论的基础上,利用Matlab具体分析了胶囊图像的特点并设计了一种破损胶囊的检测算法。该算法利用胶囊灰度图的直方图阈值分割技术,计算胶囊图像的目标面积、均值与方差。对于胶囊表面的磨损检测,利用图像的HSI颜色模型对胶囊彩色图像进行了分析与处理,磨损程度不同的胶囊在色度H分量和饱和度S分量直方图分布存在差异。以上检测方法在胶囊的小样本空间进行了实验与验证。
在软件设计方面,在熟悉整个胶囊图像处理内容的前提下,按硬件运行平台分类,详细地分析了FPGA与DSP功能实现的设计方法,以及简要的系统软件流程图。
With the great development of economy and technology, the level of people's life has consecutively improved. At the same time, people have taken care of the quality of the capsule which keeps their bodies healthy more and more rigorously. In daily life, we can catch the capsule dosage, figure, color and luster by our eyes when we buy the capsule from the drugstore. According to these physical measures, we can estimate the capsule. But in the production process of drugs in many pharmaceutical factories, this method of checking the capsule's quality one by one can not meet the need of the modern productivity of the medicine manufactory. This will not only increase the production and management costs, but also can not be guaranteed to have a high pass rate of detecting, because people's eyes have many shortcomings such as the instability and the fatigability on the capsule detection processing. Based on this background, in order to make the detection equipments be more perfect and realize the better functions, this paper predicts the running way and requirements of the existing equipment first and then make use of the digital image processing technology to do the research on the core parts which including the designing of system hardware, the algorithm of the detection of defective capsule, the designing of system software, and the simulation of the algorithm by Matlab.
For the hardware, in order to meet the requirement of running real-time of the detection equipment, this paper put forward a whole resolution scheme which is composed of DSP and FPGA on the capsule image detection system. The author studied the structure and principle of the CCD camera, the change of the video signal, the configuration of FPGA and the functions in the system, the extern interface of DSP, etc.
For the algorithm of the capsule detection, the paper introduced the basic theory and the research method of the digital image processing technology. The grey system is applied in the digital image processing, and as the base of theory in the paper. Use AGO data generation and GM(1,1) model to design a filter algorithm. On the basis of digital image processing theory, analysis the characteristic of the capsule digital image in detail by Matlab, design an algorithm on the defective capsule detection. This algorithm makes use of the segmentation technology of histogram thresholding of the capsule's grey level image to calculate the area, mean and square difference. For the detection of the defective capsule's surface, the algorithm uses the HSI color model of the images to analyse and process the color images of the capsule. The difference of the distribution of H histogram and S histogram of images can be used to classify the capsule. The detection algorithm of the capsule which introduced in this paper is proved by emulator in the space of small sample.
For the software, after knowing well about the whole image process domain, classify the running way of the system hardware and detailed analysis the functions design method of DSP and FPGA. A brief soft flow chart of system function is introduced.
在硬件方面,根据胶囊检测设备的实时性要求,本文提出了一套由DSP+FPGA组成的胶囊图像检测系统解决方案,并重点介绍了CCD摄像头的构成及工作原理、视频信号转换、FPGA配置方式及其在系统设计中的功能、DSP的外部硬件连接等相关内容。
在胶囊检测算法方面,介绍了基本的数字图像处理理论知识及常用的研究方法。本文引入了灰色系统理论作为数字图像处理的手段,利用AGO数据生成和GM(1,1)模型设计了一套数字图像滤波算法。在数字图像处理理论的基础上,利用Matlab具体分析了胶囊图像的特点并设计了一种破损胶囊的检测算法。该算法利用胶囊灰度图的直方图阈值分割技术,计算胶囊图像的目标面积、均值与方差。对于胶囊表面的磨损检测,利用图像的HSI颜色模型对胶囊彩色图像进行了分析与处理,磨损程度不同的胶囊在色度H分量和饱和度S分量直方图分布存在差异。以上检测方法在胶囊的小样本空间进行了实验与验证。
在软件设计方面,在熟悉整个胶囊图像处理内容的前提下,按硬件运行平台分类,详细地分析了FPGA与DSP功能实现的设计方法,以及简要的系统软件流程图。
With the great development of economy and technology, the level of people's life has consecutively improved. At the same time, people have taken care of the quality of the capsule which keeps their bodies healthy more and more rigorously. In daily life, we can catch the capsule dosage, figure, color and luster by our eyes when we buy the capsule from the drugstore. According to these physical measures, we can estimate the capsule. But in the production process of drugs in many pharmaceutical factories, this method of checking the capsule's quality one by one can not meet the need of the modern productivity of the medicine manufactory. This will not only increase the production and management costs, but also can not be guaranteed to have a high pass rate of detecting, because people's eyes have many shortcomings such as the instability and the fatigability on the capsule detection processing. Based on this background, in order to make the detection equipments be more perfect and realize the better functions, this paper predicts the running way and requirements of the existing equipment first and then make use of the digital image processing technology to do the research on the core parts which including the designing of system hardware, the algorithm of the detection of defective capsule, the designing of system software, and the simulation of the algorithm by Matlab.
For the hardware, in order to meet the requirement of running real-time of the detection equipment, this paper put forward a whole resolution scheme which is composed of DSP and FPGA on the capsule image detection system. The author studied the structure and principle of the CCD camera, the change of the video signal, the configuration of FPGA and the functions in the system, the extern interface of DSP, etc.
For the algorithm of the capsule detection, the paper introduced the basic theory and the research method of the digital image processing technology. The grey system is applied in the digital image processing, and as the base of theory in the paper. Use AGO data generation and GM(1,1) model to design a filter algorithm. On the basis of digital image processing theory, analysis the characteristic of the capsule digital image in detail by Matlab, design an algorithm on the defective capsule detection. This algorithm makes use of the segmentation technology of histogram thresholding of the capsule's grey level image to calculate the area, mean and square difference. For the detection of the defective capsule's surface, the algorithm uses the HSI color model of the images to analyse and process the color images of the capsule. The difference of the distribution of H histogram and S histogram of images can be used to classify the capsule. The detection algorithm of the capsule which introduced in this paper is proved by emulator in the space of small sample.
For the software, after knowing well about the whole image process domain, classify the running way of the system hardware and detailed analysis the functions design method of DSP and FPGA. A brief soft flow chart of system function is introduced.
引文
[1]刘禾.数字图像处理及应用.北京:中国电力出版社,2006
[2]张登荣.浅析药品包装的在线检测.医药包装机械,2006,2(10):44-46
[3]http://www.phmacn.com 中国制药机械网
[4]朱明,鲁剑锋.基于DSP+FPGA结构图像处理系统设计与实现.计算机测量与控制,2004,12(9):866-869
[5]王庆有.CCD应用技术.天津:天津大学出版社,2000
[6]黄慧娟,王瑞光,丁铁夫等.SAA7111A在平板显示视频接口中应用[J].液晶与显示,2004,19(4):293-297
[7]Philips Semiconductors.SAA7111A enhanced video input processor,http://cn.100y.com.tw/Pdf_file/SAA7111A HZ.pdf
[8]J.Battle,J.Marti,P Ridao.A New FPGA/DSP-Based Parallel Architecture for Real-Time Image Processing.Real-Time Imaging,2002.8(5):345-356
[9]Altera Corporation.Cyclone FPGA family data sheet.http:data.eeworld.com.cn/pdf/47198-ALTERA-EP2C5.html
[10]Deng K,Wilson J N.An approximation - based video tracking system.SPIE image Algebra and Morphlogical Image Processing,2001,11:304-312
[11]吴继华,王诚等.Altera FPGA/CPLD设计(高级篇).北京:人民邮电出版社,2005
[12]周霖.算法设计与系统方案.北京:国防工业出版社,2004
[13]汪安民,程昱,徐保根等.DSP嵌入式系统开发典型案例.北京:人民邮电出版社,2007
[14]Texas Instruments.TMS320C6711,TMS320C6711B FLOATING-POINT SIGNAL PROCESSORS,2004
[15]任丽香,马淑芬,李芳慧等.TMS320C6000系列DSPs的原理与应用.北京:电子工业出版社,2000
[16]Silicon Storage Technology.16Mbit(×16)Multi-purpose Flash SST39LF160/SST39VF160,2003
[17]王念旭.DSP基础与应用系统设计.北京:北京航空航天大学出版社,2001
[18]贾云德.机器视觉.北京:科学出版社,2000
[19]马颂德,张正友等.计算机视觉—计算理论与算法基础.北京:科学出版社,1996
[20]徐建华.图像处理与分析.北京:科学出版社,2002
[21]章毓晋.图像工程(上册)-图像处理和分析.北京:清华大学出版社,2000
[22]张兆礼.现代图像处理技术及Matlab实现.北京:人民邮电出版社,2001
[23]江景涛,姜学东,李福荣等.利用中值滤波去除图像噪声的研究及Matlab实现.莱阳农学院学报,2006,23(1):63-65
[24]熊和金,徐华中.灰色控制.北京:国防工业出版社,2005
[25]邓聚龙.灰色预测与决策.武汉:华中工学院出版社,1985
[26]keneth R.Castleman.数字图像处理.北京:电子工业出版社,2002
[27]章毓晋.图像分割.北京:科学出版社,2000
[28]朱志刚,石定机等.数字图像处理.北京:电子工业出版社,1998
[29]刘超,钱光弟.用FPGA实现嵌入式视频图像信号实时采集.实验科学与技术,2005,6(02):12-15
[30]丁宝明.高速数字视频实时采集系统研究:[硕士学位论文].南京:南京理工大学,2006
[31]朱鹏飞,赵雅兴.视频图像捕获系统SRAM控制器的FPGA实现.半导体技术,2002,6(27):18-21
[32]居里锴.计算机视觉实时图像处理检测系统的研究与设计:[硕士学文论文].南京:南京理工大学,2004
[33]张雅绮,李锵等.Verilog HDL高级数字设计.北京:电子工业出版社,2005
[34]梁吉,蒋式勤等.视觉检测系统及应用.微计算机信息,2003,8(12):63-66
[35]李立轻,黄秀宝.图像处理用于玻璃疵点自动检测的研究进展.东华大学学报,2002,28(4):118-122
[36]National Semiconductor Corp.LVDS Owner's manual.National semiconductor Corp Revision 2.0,2000:24-123
[37]R Halverson,Art Lew.FPGAs for expression level parallel processing.Microprocessors and Microsystems,1995:533-540
[38]Bruce Draper,Walid Najjar,Wim Bohm.Compiling and optimizing image processing algorithms for FPGA's Workshop on Computer.Architecture for Machine Performance,2000.222-231
[39]Samir Tagzout,Karim Achour,Oualid Djekoune,Hough transform algorithm for FPGA implementation.Signal Processing,2001(81),1295-1301
[40]王德生,徐婉莹,黄新生等.基于FPGA的图像预处理快速算法及仿真.计算机仿真,2007,24(08):320-322
[2]张登荣.浅析药品包装的在线检测.医药包装机械,2006,2(10):44-46
[3]http://www.phmacn.com 中国制药机械网
[4]朱明,鲁剑锋.基于DSP+FPGA结构图像处理系统设计与实现.计算机测量与控制,2004,12(9):866-869
[5]王庆有.CCD应用技术.天津:天津大学出版社,2000
[6]黄慧娟,王瑞光,丁铁夫等.SAA7111A在平板显示视频接口中应用[J].液晶与显示,2004,19(4):293-297
[7]Philips Semiconductors.SAA7111A enhanced video input processor,http://cn.100y.com.tw/Pdf_file/SAA7111A HZ.pdf
[8]J.Battle,J.Marti,P Ridao.A New FPGA/DSP-Based Parallel Architecture for Real-Time Image Processing.Real-Time Imaging,2002.8(5):345-356
[9]Altera Corporation.Cyclone FPGA family data sheet.http:data.eeworld.com.cn/pdf/47198-ALTERA-EP2C5.html
[10]Deng K,Wilson J N.An approximation - based video tracking system.SPIE image Algebra and Morphlogical Image Processing,2001,11:304-312
[11]吴继华,王诚等.Altera FPGA/CPLD设计(高级篇).北京:人民邮电出版社,2005
[12]周霖.算法设计与系统方案.北京:国防工业出版社,2004
[13]汪安民,程昱,徐保根等.DSP嵌入式系统开发典型案例.北京:人民邮电出版社,2007
[14]Texas Instruments.TMS320C6711,TMS320C6711B FLOATING-POINT SIGNAL PROCESSORS,2004
[15]任丽香,马淑芬,李芳慧等.TMS320C6000系列DSPs的原理与应用.北京:电子工业出版社,2000
[16]Silicon Storage Technology.16Mbit(×16)Multi-purpose Flash SST39LF160/SST39VF160,2003
[17]王念旭.DSP基础与应用系统设计.北京:北京航空航天大学出版社,2001
[18]贾云德.机器视觉.北京:科学出版社,2000
[19]马颂德,张正友等.计算机视觉—计算理论与算法基础.北京:科学出版社,1996
[20]徐建华.图像处理与分析.北京:科学出版社,2002
[21]章毓晋.图像工程(上册)-图像处理和分析.北京:清华大学出版社,2000
[22]张兆礼.现代图像处理技术及Matlab实现.北京:人民邮电出版社,2001
[23]江景涛,姜学东,李福荣等.利用中值滤波去除图像噪声的研究及Matlab实现.莱阳农学院学报,2006,23(1):63-65
[24]熊和金,徐华中.灰色控制.北京:国防工业出版社,2005
[25]邓聚龙.灰色预测与决策.武汉:华中工学院出版社,1985
[26]keneth R.Castleman.数字图像处理.北京:电子工业出版社,2002
[27]章毓晋.图像分割.北京:科学出版社,2000
[28]朱志刚,石定机等.数字图像处理.北京:电子工业出版社,1998
[29]刘超,钱光弟.用FPGA实现嵌入式视频图像信号实时采集.实验科学与技术,2005,6(02):12-15
[30]丁宝明.高速数字视频实时采集系统研究:[硕士学位论文].南京:南京理工大学,2006
[31]朱鹏飞,赵雅兴.视频图像捕获系统SRAM控制器的FPGA实现.半导体技术,2002,6(27):18-21
[32]居里锴.计算机视觉实时图像处理检测系统的研究与设计:[硕士学文论文].南京:南京理工大学,2004
[33]张雅绮,李锵等.Verilog HDL高级数字设计.北京:电子工业出版社,2005
[34]梁吉,蒋式勤等.视觉检测系统及应用.微计算机信息,2003,8(12):63-66
[35]李立轻,黄秀宝.图像处理用于玻璃疵点自动检测的研究进展.东华大学学报,2002,28(4):118-122
[36]National Semiconductor Corp.LVDS Owner's manual.National semiconductor Corp Revision 2.0,2000:24-123
[37]R Halverson,Art Lew.FPGAs for expression level parallel processing.Microprocessors and Microsystems,1995:533-540
[38]Bruce Draper,Walid Najjar,Wim Bohm.Compiling and optimizing image processing algorithms for FPGA's Workshop on Computer.Architecture for Machine Performance,2000.222-231
[39]Samir Tagzout,Karim Achour,Oualid Djekoune,Hough transform algorithm for FPGA implementation.Signal Processing,2001(81),1295-1301
[40]王德生,徐婉莹,黄新生等.基于FPGA的图像预处理快速算法及仿真.计算机仿真,2007,24(08):320-322