仪器仪表数字图像识别及其应用研究
|
摘要
仪器仪表被广泛地应用于各个行业中测量系统中,由于某些仪器仪表并没有提供数据传送的接口,要实现自动化,需要利用摄像技术、图像处理技术、图像识别技术以及通信技术,来实现仪器仪表读数识别具有的实用价值。
本文首先分析了数字识别的现状以及发展趋势,针对数字识别系统的研究,介绍了系统的硬件和软件平台,有利于数字识别系统的应用研究。在图像采集方面,直接采用USB摄像头,省去采集卡,降低了成本;针对体温计图像,采取了合理的图像增强、图像去噪、图像分割等预处理技术,以利于后续的处理。采用图像梯度结合先验知识来分割图像,取得了良好的效果;在图像二值化方面,采用了基于边缘检测和种子填充的二值化分割方法,对canny算子、种子填充给予了详细的介绍,并给出了处理结果,这种二值化分割方法分割准确、适应性强,对于较暗、较亮,及反色图像都能较准确的分割;采用BP神经网络实现数字识别,提高了识别率。
本文针对的是没有数字输出接口的体温计显示数字图像,给出了实现机器视觉识别的流程图,以实现智能马桶的应用中的体温计数据的自动采集,具有实际应用价值。在体温计显示数字图像预处理、分割、二值化方面具有特色。
最后,对本文的工作进行了总结,提出了存在的问题和进一步改进的方向。
Instrument is used in measure system widely in many industries, as for some instruments that have no interface to send data, in order to automation, image sampling technique, image processing technique, image recognition technique and communication are required. So instrument recognition is very valuable.This thesis analyzes the current condition and development trend of digit recognition firstly, introducing the platform of hardware and software, facilitate the application research of digit recognition system. Image getting uses USB camera directly, cut down the cost in some degree. Some measures are taken to preprocess thermometer image, such as image enhance, image denoising, image segmentation and so on. Image gradient and previous experience technology are taken to segment image, the method is very effective. Edge detection and seed filling algorithm are taken to binarize the image, introducing canny operator and seed filling. This binarization method is precise and flexible, not only to normal image, but also to brighter and darker image. BP neural network is used to recognize digit, this improves the recognition rate.The object of this thesis is the digital image of thermometer having no data output interface and used in intelligent toilet, processing flowchart and steps are given, accomplishing data automatic sampling, having very high application value. In the digital image sampling, segmentation and binarization there is own characteristic.In the end of this thesis, concluding the total contents, pointing out the current problems to be resolved.
本文首先分析了数字识别的现状以及发展趋势,针对数字识别系统的研究,介绍了系统的硬件和软件平台,有利于数字识别系统的应用研究。在图像采集方面,直接采用USB摄像头,省去采集卡,降低了成本;针对体温计图像,采取了合理的图像增强、图像去噪、图像分割等预处理技术,以利于后续的处理。采用图像梯度结合先验知识来分割图像,取得了良好的效果;在图像二值化方面,采用了基于边缘检测和种子填充的二值化分割方法,对canny算子、种子填充给予了详细的介绍,并给出了处理结果,这种二值化分割方法分割准确、适应性强,对于较暗、较亮,及反色图像都能较准确的分割;采用BP神经网络实现数字识别,提高了识别率。
本文针对的是没有数字输出接口的体温计显示数字图像,给出了实现机器视觉识别的流程图,以实现智能马桶的应用中的体温计数据的自动采集,具有实际应用价值。在体温计显示数字图像预处理、分割、二值化方面具有特色。
最后,对本文的工作进行了总结,提出了存在的问题和进一步改进的方向。
Instrument is used in measure system widely in many industries, as for some instruments that have no interface to send data, in order to automation, image sampling technique, image processing technique, image recognition technique and communication are required. So instrument recognition is very valuable.This thesis analyzes the current condition and development trend of digit recognition firstly, introducing the platform of hardware and software, facilitate the application research of digit recognition system. Image getting uses USB camera directly, cut down the cost in some degree. Some measures are taken to preprocess thermometer image, such as image enhance, image denoising, image segmentation and so on. Image gradient and previous experience technology are taken to segment image, the method is very effective. Edge detection and seed filling algorithm are taken to binarize the image, introducing canny operator and seed filling. This binarization method is precise and flexible, not only to normal image, but also to brighter and darker image. BP neural network is used to recognize digit, this improves the recognition rate.The object of this thesis is the digital image of thermometer having no data output interface and used in intelligent toilet, processing flowchart and steps are given, accomplishing data automatic sampling, having very high application value. In the digital image sampling, segmentation and binarization there is own characteristic.In the end of this thesis, concluding the total contents, pointing out the current problems to be resolved.
引文
[1] 何斌,马天予.Visual C++数字图像处理[M].人民邮电出版社:2002年12月.
[2] 夏良正.数字图像处理[M].东南大学出版社:1999年9月.
[3] 苏严华.Visual C++数字图像识别技术典型案例[M].人民邮电出版社:2004年8月.
[4] 边肇祺,张学工.模式识别[M].清华大学出版社:1999年8月.
[5] 关山,王伟杰.基于BP神经网络的车牌字符自动识别方法的研究[J].东北电力学院学报,2004 24(4):56-60.
[6] 付忠良.一种新的图像阈值选取方法[J].计算机应用,2003.20(10):13-15.
[7] 王爱民,沈兰荪.图像分割研究综述[J].测控技术,2000.19(5):1-6.
[8] 章琉晋.图象处理和分析[M].北京清华大学出版社.1999年3月
[9] 林卉,赵长胜.基于Canny算子的边缘检测及评价[J].黑龙江工程学陇学报.2003.17(2):3-7.
[10] 刘相滨,胡峰松.一种新的区域种子填充算法[J].计算机工程与应用,2002.38(8):100-102.
[11] 胡英,陈辉.基于分形与投影法的汽车牌照定位[J].河北工业大学学报,2004.23(5):56-59.
[12] 张玉姣,史忠科.基于连通体检测及投影法的牌照字符切分[J].小型微型计算机系统,2004.25(4):564-567.
[13] 董立菊.图像阈值化技术的综述、分类及评价[J].沈阳大学学报,2004.4(1):8-11.
[14] 韩思奇,王蕾.图像分割的阈值法综述[J].系统工程与电子技术,2002.24(6):91-96.
[15] 李了了,邓善熙.基于大津法的图像分块二值化算法[J].微计算机信息,2005.21(8-3):76-79.
[16] 王三武,戴亚文,景仁坤.复杂仪表的图像自动读数识别方法.武汉理工大学学报,2003,6(25):76-78.
[17] 张海波,段会川,张曙光等.一种数字仪表显示值快速识别方法.计算机工程与应用,2005.41(4):223-226.
[18] 张重阳,娄震.基于轮廓和统计特征的手写体数字识别[J].计算机工程与应用,2004.40(9):83-87.
[19] 李根强.使用边缘检测技术实现字符轮廓线的提取及显示[J].计算技术与自动化,2000.19(2):66-69.
[20] 焦李成.神经网络系统理论[M].西安电子科技大学出版社:1990年.
[21] Nikhil R. Pal and Sankar K. Pal. A Review on Image Segmentation Techniques[J]. Pattern Recognition. 1999. 26(9): 1272-1294.
[22] Demuth H d, Beale M. neural network toolbox use's guide Version 4[M]. the mathworks Inc, 2000.
[23] Char F H Y, Lain F K, Zhu H. Adaptive Thresholdingby Vaiiational Method[J]. IEEE Trans Image Processing, 1998. 7(3): 468-473.
[24] P. K. Sahoo and A. K. C. Wong. A Survery of Thresholding Techniques[J]. Graphics and Image Proceeding, 1998. 41: 233-266.
[25] F. Martin Rodriguez et al, New advances in Automatic Reading of VLP's [J]. Proc. SPC-2000(IASTED), Marbella. spain, 2000: 126-131.
[26] J. A. G Nijhuis. Car License Plate Recognition with Neural Networks and Fuzzy Logic[J]. IEEE 1995: 123-128
[27] Bernsen. Proc. of 80th Int'l Conf [J]. On Part. Recogn. 1986: 1251-1255.
[28] M. A. Arregui, J. A. Mitxelena. Automatic recognition of number plates[J].the Ivth Spanish Symposium, 1992.1(1): 239-258.
[29] Yong Sung Soh. Design of Real Time Vehicle Identification System[J]. IEEE, 1994. 4: 235-240
[30] Paolo COmelli, Paolo Ferragina. Optical Recognition of Motor Vehicle License Plates[J]. IEEE Trans on Vehicle Technology, 44(4): 163-172.
[31] 李江华,谢红,王晓丹.基于VFW实时视频捕获原理与实现[J].应用科技,2005.32(9):53-55.
[32] 雷丰中,谈新权,雷海军.USB2.0视频图像采集系统设计[J].湖南工程学院学报,2005.15(2):1-3.
[33] 唐苏湘,汪仁煌.网络化智能马桶的研究[J].自动化技术与应用,2005.24(6):78-80.
[34] 时道波.Premiere中常用12种图像文什格式简介[J].中国有线电视,2005(22):2232-2233.
[35] 李辉,徐琦,张安,王安丽.VC++串行通信编程方法研究[J].微处理机,200.26(3):36-39.
[36] 刘伟.在VC中利用ActiveX控件实现串行通信[J].现代通信技术,2003(4):66-69.
[37] 黄艺,杜宇人.基于边缘信息的图像分割技术研究[J].现代电子技术.2005.28(5):116-117.
[38] De Natale, Francesco G. B. Edge detection by point classification of Canny filtered images[J]. Signal Processing, 1997. 60(1): 11-22.
[39] Schubert, Li Hui. Chinese word segmentation and its effect on information retrieval [J]. Information Processing and Management, 2004. 40(1): 161-190.
[40] Zheng Danian, Zhao Yannan, Wang Jiaxin. An efficient method of license plate location[J].Pattern Recognition Letters, 2005. 26(5):2431"2438.
[41] Yin Peng Yeng. A fast scheme for optimal thresholding using genetic algorithms[J].Signal Processing, 1999. 72(2):85-95.
[2] 夏良正.数字图像处理[M].东南大学出版社:1999年9月.
[3] 苏严华.Visual C++数字图像识别技术典型案例[M].人民邮电出版社:2004年8月.
[4] 边肇祺,张学工.模式识别[M].清华大学出版社:1999年8月.
[5] 关山,王伟杰.基于BP神经网络的车牌字符自动识别方法的研究[J].东北电力学院学报,2004 24(4):56-60.
[6] 付忠良.一种新的图像阈值选取方法[J].计算机应用,2003.20(10):13-15.
[7] 王爱民,沈兰荪.图像分割研究综述[J].测控技术,2000.19(5):1-6.
[8] 章琉晋.图象处理和分析[M].北京清华大学出版社.1999年3月
[9] 林卉,赵长胜.基于Canny算子的边缘检测及评价[J].黑龙江工程学陇学报.2003.17(2):3-7.
[10] 刘相滨,胡峰松.一种新的区域种子填充算法[J].计算机工程与应用,2002.38(8):100-102.
[11] 胡英,陈辉.基于分形与投影法的汽车牌照定位[J].河北工业大学学报,2004.23(5):56-59.
[12] 张玉姣,史忠科.基于连通体检测及投影法的牌照字符切分[J].小型微型计算机系统,2004.25(4):564-567.
[13] 董立菊.图像阈值化技术的综述、分类及评价[J].沈阳大学学报,2004.4(1):8-11.
[14] 韩思奇,王蕾.图像分割的阈值法综述[J].系统工程与电子技术,2002.24(6):91-96.
[15] 李了了,邓善熙.基于大津法的图像分块二值化算法[J].微计算机信息,2005.21(8-3):76-79.
[16] 王三武,戴亚文,景仁坤.复杂仪表的图像自动读数识别方法.武汉理工大学学报,2003,6(25):76-78.
[17] 张海波,段会川,张曙光等.一种数字仪表显示值快速识别方法.计算机工程与应用,2005.41(4):223-226.
[18] 张重阳,娄震.基于轮廓和统计特征的手写体数字识别[J].计算机工程与应用,2004.40(9):83-87.
[19] 李根强.使用边缘检测技术实现字符轮廓线的提取及显示[J].计算技术与自动化,2000.19(2):66-69.
[20] 焦李成.神经网络系统理论[M].西安电子科技大学出版社:1990年.
[21] Nikhil R. Pal and Sankar K. Pal. A Review on Image Segmentation Techniques[J]. Pattern Recognition. 1999. 26(9): 1272-1294.
[22] Demuth H d, Beale M. neural network toolbox use's guide Version 4[M]. the mathworks Inc, 2000.
[23] Char F H Y, Lain F K, Zhu H. Adaptive Thresholdingby Vaiiational Method[J]. IEEE Trans Image Processing, 1998. 7(3): 468-473.
[24] P. K. Sahoo and A. K. C. Wong. A Survery of Thresholding Techniques[J]. Graphics and Image Proceeding, 1998. 41: 233-266.
[25] F. Martin Rodriguez et al, New advances in Automatic Reading of VLP's [J]. Proc. SPC-2000(IASTED), Marbella. spain, 2000: 126-131.
[26] J. A. G Nijhuis. Car License Plate Recognition with Neural Networks and Fuzzy Logic[J]. IEEE 1995: 123-128
[27] Bernsen. Proc. of 80th Int'l Conf [J]. On Part. Recogn. 1986: 1251-1255.
[28] M. A. Arregui, J. A. Mitxelena. Automatic recognition of number plates[J].the Ivth Spanish Symposium, 1992.1(1): 239-258.
[29] Yong Sung Soh. Design of Real Time Vehicle Identification System[J]. IEEE, 1994. 4: 235-240
[30] Paolo COmelli, Paolo Ferragina. Optical Recognition of Motor Vehicle License Plates[J]. IEEE Trans on Vehicle Technology, 44(4): 163-172.
[31] 李江华,谢红,王晓丹.基于VFW实时视频捕获原理与实现[J].应用科技,2005.32(9):53-55.
[32] 雷丰中,谈新权,雷海军.USB2.0视频图像采集系统设计[J].湖南工程学院学报,2005.15(2):1-3.
[33] 唐苏湘,汪仁煌.网络化智能马桶的研究[J].自动化技术与应用,2005.24(6):78-80.
[34] 时道波.Premiere中常用12种图像文什格式简介[J].中国有线电视,2005(22):2232-2233.
[35] 李辉,徐琦,张安,王安丽.VC++串行通信编程方法研究[J].微处理机,200.26(3):36-39.
[36] 刘伟.在VC中利用ActiveX控件实现串行通信[J].现代通信技术,2003(4):66-69.
[37] 黄艺,杜宇人.基于边缘信息的图像分割技术研究[J].现代电子技术.2005.28(5):116-117.
[38] De Natale, Francesco G. B. Edge detection by point classification of Canny filtered images[J]. Signal Processing, 1997. 60(1): 11-22.
[39] Schubert, Li Hui. Chinese word segmentation and its effect on information retrieval [J]. Information Processing and Management, 2004. 40(1): 161-190.
[40] Zheng Danian, Zhao Yannan, Wang Jiaxin. An efficient method of license plate location[J].Pattern Recognition Letters, 2005. 26(5):2431"2438.
[41] Yin Peng Yeng. A fast scheme for optimal thresholding using genetic algorithms[J].Signal Processing, 1999. 72(2):85-95.