基于图像处理的散货船港航交重计量系统
|
摘要
水尺计重(Draft survey)是测定船舶装、卸货物前后的平均吃水数据所对应的排水量差值,扣除船舶中的其它载荷重量计算船舶载货量的方法。煤炭、矿石、化肥等大宗散货物品通常用此方法进行计量。
本课题主要研究了利用图像处理技术确定人眼不易识别的吃水线位置。针对水尺图像采集环境的特点,本课题采用了基于爬壁机器人为载体的图像采集装置；针对水尺图像的特点,本课题采用了基于形态学、投影和图像分割的算法对水尺图像进行的检测处理。
基于提高后续检测精度的目的,系统采用高分辨率的图像采集装置进行水尺图像的采集。为了降低后续处理的数据量,首先对图像进行二值化分割,获得高亮度的字符区域,并以此为基础在高度方向对图像进行投影统计,提取出离水线最近的“M”字符区域并进行倾斜校正,然后对”M”字符区域进行分割和对“M”字符前面的数字进行识别。对“M”下面的小字符进行分割和识别,并确定离水面最近的可识别字符。为了防止因船舶形状对字符大小的影响,可以根据数字的识别效果以“M”字符所占的像素个数或者离水线最近的可识别数字所占的像素个数为计量标准。再次以可识别数字的下边缘为界对彩色图像进行剪裁,然后对彩色图像进行分割以及提取最长边界并拟合水面位置,为了防止检测结果受海水波浪的影响,对检测结果进行均值运算确定水面所在的位置。
本文针对船舶水尺字符的特点,对船舶水尺图像进行了充分的预处理,包括对字符的分割、倾斜校正和图像剪裁,获得了较好的检测和识别效果；通过彩色图像分割确定吃水线位置,按照水尺计重的规则,利用“M”前面的字符识别结果、计量标准和水面位置计算水尺数据。通过对实地采集的图像进行处理检测,表明此方法是有效的,实现船舶的自动水尺计量系统是可行的。
Draft survey is a method of through the determination of ship's average draft corresponding to the difference in value of displacement before and after unloading,deduct other load weight and calculate the amount of dead weight.The method is usually as the measurement of the major bulks,such as coal,ore,fertilizer and so on.
The paper mainly does the research on how to get the position of the waterline using image processing technology,which is not easy for human eye. According to the characteristics of draft image acquisition,the paper adopts climbing robot as the carrier of image acquisition device.The paper adopts algorithm processing draft image based on morphological,projection and image segmentation.
In order to improve the measurement precision,using the high resolution image acquisition device.For reducing the volume of the following-up processing,the paper firstly take the binary segmentation to get the characters area of high brightness,and projection statistics in the height of image direction,then extract the recent "M"character area from waterline and tilt correction,next segment the "M"regional and identify the digital before the "M"Segmentation and recognition the small character below the "M"and determined the character which is nearest from the waterline.In order to prevent the shape of ship influence the size of characters,we could use the number of pixels for measure standers,the "M"characters or the digital which is nearest from the waterline and recognition.Clipping the color image with a boundary that is the bottom edge of identified digital.And then segmenting the color image and extracting the longest boundary and fitting the waterline position.
According to the characteristics of the draft character,we carried out a full image pretreatment process, including the character segmentation,tilt correction and image clipping,whichhas good processing and recognition effect.Then using the color image segmentation determine the position of the waterline. Finally, calculating draft data according to the recognition results of the digital before the "M" area, measurement standard and the position of waterline using the draft survey regulations.
After the multiple experiments on many images, it is proved that this method is effective and realizing the ship's automatic draft survey system is feasible.
本课题主要研究了利用图像处理技术确定人眼不易识别的吃水线位置。针对水尺图像采集环境的特点,本课题采用了基于爬壁机器人为载体的图像采集装置；针对水尺图像的特点,本课题采用了基于形态学、投影和图像分割的算法对水尺图像进行的检测处理。
基于提高后续检测精度的目的,系统采用高分辨率的图像采集装置进行水尺图像的采集。为了降低后续处理的数据量,首先对图像进行二值化分割,获得高亮度的字符区域,并以此为基础在高度方向对图像进行投影统计,提取出离水线最近的“M”字符区域并进行倾斜校正,然后对”M”字符区域进行分割和对“M”字符前面的数字进行识别。对“M”下面的小字符进行分割和识别,并确定离水面最近的可识别字符。为了防止因船舶形状对字符大小的影响,可以根据数字的识别效果以“M”字符所占的像素个数或者离水线最近的可识别数字所占的像素个数为计量标准。再次以可识别数字的下边缘为界对彩色图像进行剪裁,然后对彩色图像进行分割以及提取最长边界并拟合水面位置,为了防止检测结果受海水波浪的影响,对检测结果进行均值运算确定水面所在的位置。
本文针对船舶水尺字符的特点,对船舶水尺图像进行了充分的预处理,包括对字符的分割、倾斜校正和图像剪裁,获得了较好的检测和识别效果；通过彩色图像分割确定吃水线位置,按照水尺计重的规则,利用“M”前面的字符识别结果、计量标准和水面位置计算水尺数据。通过对实地采集的图像进行处理检测,表明此方法是有效的,实现船舶的自动水尺计量系统是可行的。
Draft survey is a method of through the determination of ship's average draft corresponding to the difference in value of displacement before and after unloading,deduct other load weight and calculate the amount of dead weight.The method is usually as the measurement of the major bulks,such as coal,ore,fertilizer and so on.
The paper mainly does the research on how to get the position of the waterline using image processing technology,which is not easy for human eye. According to the characteristics of draft image acquisition,the paper adopts climbing robot as the carrier of image acquisition device.The paper adopts algorithm processing draft image based on morphological,projection and image segmentation.
In order to improve the measurement precision,using the high resolution image acquisition device.For reducing the volume of the following-up processing,the paper firstly take the binary segmentation to get the characters area of high brightness,and projection statistics in the height of image direction,then extract the recent "M"character area from waterline and tilt correction,next segment the "M"regional and identify the digital before the "M"Segmentation and recognition the small character below the "M"and determined the character which is nearest from the waterline.In order to prevent the shape of ship influence the size of characters,we could use the number of pixels for measure standers,the "M"characters or the digital which is nearest from the waterline and recognition.Clipping the color image with a boundary that is the bottom edge of identified digital.And then segmenting the color image and extracting the longest boundary and fitting the waterline position.
According to the characteristics of the draft character,we carried out a full image pretreatment process, including the character segmentation,tilt correction and image clipping,whichhas good processing and recognition effect.Then using the color image segmentation determine the position of the waterline. Finally, calculating draft data according to the recognition results of the digital before the "M" area, measurement standard and the position of waterline using the draft survey regulations.
After the multiple experiments on many images, it is proved that this method is effective and realizing the ship's automatic draft survey system is feasible.
引文
[1]周广程.图像处理技术在船舶吃水自动检测系统中的应用：(硕士学位论文).南京：南京理工大学,2006.
[2]刘仁金,高远飙,郝祥根.船舶吃水线定位分析及算法研究.皖西学报.2009,25(5)1-4.
[3]孙国元,毛奇凰.自动检测船舶吃水和稳性参数的方法探讨.中国航海.2002,51(2)28-30.
[4]刘辉强.水尺计重及其误差分析：(硕士学位论文).大连海事大学,2010.6,1-40.
[5]龙钧宇,金连文.一种基于全局均值和局部方差的图像二值化方法.计算机工程.2004(2).
[6]马桂珍,房宗良,姚宗中.SUSAN边缘检测算法性能分析与比较.现代电子技术.2007(8).
[7]贡丽霞,白艳萍.Radon变换在倾斜车牌图像校正中的应用.测试技术学报.2009(23).
[8]袁凤刚,刘建成.不同插值方法实现数字图像旋转研究.软件导刊.2010(4).
[9]曹佃国,陈浩杰,李鹏.基于Matlab的双线性插值算法在图像旋转中的应用.中国印刷与包装研究.2010(4).
[10]范立南,韩晓微,张广渊.图像处理与模式识别.科学出版社,2007.
[11]Mark S. Nixon, Alberto S. Aguado. Feature Extraction and Image Processing. BeiJing:Publishing house of electronics industry.2010.
[12]常娜.图像处理中的边缘检测算法研究综述.中国科技信息.2011(4).
[13]魏宝刚,李向阳,鲁东明,潘云鹤.彩色图像分割研究进展.计算机科学.1999(4).
[14]岗萨雷斯.数字图像处理(MATLAB版)[M].阮秋琦,译.北京：电子工业出版社,2005.
[15]张钢.散装货物运输中水尺计重的原则和方法[J].中国航海,2006,36-40.
[16]陈亚飞,汪益兵.海运固体散装货物的水尺计重[J],航海技术,2010,3,36-38.
[17]朱永俊,鲍宏杨.船舶水尺计重误差的因素分析[J],南通航运职业技术学报,2008,4,57-59.
[18]王建平.船舶载重性能的应用[J],世界海运,1995,1.
[19]郭方.基于视频的船舶吃水线检测方法的研究[D].大连海事大学,2010：9-36.
[20]严国莉,黄山,李岱璋,尚建华.印刷体数字快速识别算法在身份证编号数字识别中的应用[J].计算机工程,2003,29(1).
[21]朱铭煜,周武能.图像处理在药片检测中的应用[J].微型电脑应用,2010,26(5).
[22]戚尚菊,纪秀花,基于边缘的结构相似度模糊图像质量评价[J].计算机工程与科学,2011, 33(2).
[23]郑秋梅,王红霞,闵利田.基于内外边缘颜色特征的图像检索算法[J].工程图学学报,2010,2.
[24]常娜.图像处理中的边缘检测算法研究综述[J].中国科技信息,2011,4.
[25]许捍卫,王成.一种简单的数字识别方法研究[J].2000.
[26]王晓军.不变矩在图像特征提取及目标识别中的应用[J].2011,1.
[27]Rafael C.Gonzalez,Richard E.Woods数字图像处理.第2版.北京：电子工业出版社.2003.
[28]Otsu N.A threshold selection method from gray-level histogram[J],IEEE Trans Action SMC,1979(9):652-655.
[29]李文举,梁德群,张旗,王演.基于方向区域距离测度的彩色边缘检测方法.计算机应用.Vo1.23,No.1,2003.
[30]连静,于珂.基于多尺度融合技术的图像边缘检测.仪器仪表学报.Vo1.28,No.5,2007.
[31]HAN Junwei,GUO Lei,An Application of Stochastic Heuristic Search Method to Edge Extraction in Noisy Image[A],Proc.SPIE 2001,Vol.4550:57-62.
[32]Eichel P H,Delp E J.Sequential edge linking.In:Proc:22nd Allerton Conf Comun[A].Control and Computers[C].Monticello:1984,782-791.
[33]张卫红.图像边缘提取的启发式搜索算法.航空计算技术.Vo1.35 No.22005(6)：42-44.
[34]沈俊.边界探测最佳线性算子[J].模式识别与人工智能.1987(1)：86-103.
[35]雷丽珍.数字图像边缘检测方法的探讨.测绘通报.2006(3)：40-41.
[36]马志锋,杨水超,赵保军,何佩琨.改进的快速图像模糊边缘检测算法.激光与红外.Vo1.35No.42005(4)：300-302.
[37]于烨,陆建华,郑君里.一种新的彩色图像边缘检测算法.清华大学学报(自然科学版).Vo1.45No.102005(10)：1339-1343.
[38]杨平先,孙兴波.一种改进的多尺度形态边缘检测算法.光电工程.Vo1.32 No.112005(11)：72-75.
[2]刘仁金,高远飙,郝祥根.船舶吃水线定位分析及算法研究.皖西学报.2009,25(5)1-4.
[3]孙国元,毛奇凰.自动检测船舶吃水和稳性参数的方法探讨.中国航海.2002,51(2)28-30.
[4]刘辉强.水尺计重及其误差分析：(硕士学位论文).大连海事大学,2010.6,1-40.
[5]龙钧宇,金连文.一种基于全局均值和局部方差的图像二值化方法.计算机工程.2004(2).
[6]马桂珍,房宗良,姚宗中.SUSAN边缘检测算法性能分析与比较.现代电子技术.2007(8).
[7]贡丽霞,白艳萍.Radon变换在倾斜车牌图像校正中的应用.测试技术学报.2009(23).
[8]袁凤刚,刘建成.不同插值方法实现数字图像旋转研究.软件导刊.2010(4).
[9]曹佃国,陈浩杰,李鹏.基于Matlab的双线性插值算法在图像旋转中的应用.中国印刷与包装研究.2010(4).
[10]范立南,韩晓微,张广渊.图像处理与模式识别.科学出版社,2007.
[11]Mark S. Nixon, Alberto S. Aguado. Feature Extraction and Image Processing. BeiJing:Publishing house of electronics industry.2010.
[12]常娜.图像处理中的边缘检测算法研究综述.中国科技信息.2011(4).
[13]魏宝刚,李向阳,鲁东明,潘云鹤.彩色图像分割研究进展.计算机科学.1999(4).
[14]岗萨雷斯.数字图像处理(MATLAB版)[M].阮秋琦,译.北京：电子工业出版社,2005.
[15]张钢.散装货物运输中水尺计重的原则和方法[J].中国航海,2006,36-40.
[16]陈亚飞,汪益兵.海运固体散装货物的水尺计重[J],航海技术,2010,3,36-38.
[17]朱永俊,鲍宏杨.船舶水尺计重误差的因素分析[J],南通航运职业技术学报,2008,4,57-59.
[18]王建平.船舶载重性能的应用[J],世界海运,1995,1.
[19]郭方.基于视频的船舶吃水线检测方法的研究[D].大连海事大学,2010：9-36.
[20]严国莉,黄山,李岱璋,尚建华.印刷体数字快速识别算法在身份证编号数字识别中的应用[J].计算机工程,2003,29(1).
[21]朱铭煜,周武能.图像处理在药片检测中的应用[J].微型电脑应用,2010,26(5).
[22]戚尚菊,纪秀花,基于边缘的结构相似度模糊图像质量评价[J].计算机工程与科学,2011, 33(2).
[23]郑秋梅,王红霞,闵利田.基于内外边缘颜色特征的图像检索算法[J].工程图学学报,2010,2.
[24]常娜.图像处理中的边缘检测算法研究综述[J].中国科技信息,2011,4.
[25]许捍卫,王成.一种简单的数字识别方法研究[J].2000.
[26]王晓军.不变矩在图像特征提取及目标识别中的应用[J].2011,1.
[27]Rafael C.Gonzalez,Richard E.Woods数字图像处理.第2版.北京：电子工业出版社.2003.
[28]Otsu N.A threshold selection method from gray-level histogram[J],IEEE Trans Action SMC,1979(9):652-655.
[29]李文举,梁德群,张旗,王演.基于方向区域距离测度的彩色边缘检测方法.计算机应用.Vo1.23,No.1,2003.
[30]连静,于珂.基于多尺度融合技术的图像边缘检测.仪器仪表学报.Vo1.28,No.5,2007.
[31]HAN Junwei,GUO Lei,An Application of Stochastic Heuristic Search Method to Edge Extraction in Noisy Image[A],Proc.SPIE 2001,Vol.4550:57-62.
[32]Eichel P H,Delp E J.Sequential edge linking.In:Proc:22nd Allerton Conf Comun[A].Control and Computers[C].Monticello:1984,782-791.
[33]张卫红.图像边缘提取的启发式搜索算法.航空计算技术.Vo1.35 No.22005(6)：42-44.
[34]沈俊.边界探测最佳线性算子[J].模式识别与人工智能.1987(1)：86-103.
[35]雷丽珍.数字图像边缘检测方法的探讨.测绘通报.2006(3)：40-41.
[36]马志锋,杨水超,赵保军,何佩琨.改进的快速图像模糊边缘检测算法.激光与红外.Vo1.35No.42005(4)：300-302.
[37]于烨,陆建华,郑君里.一种新的彩色图像边缘检测算法.清华大学学报(自然科学版).Vo1.45No.102005(10)：1339-1343.
[38]杨平先,孙兴波.一种改进的多尺度形态边缘检测算法.光电工程.Vo1.32 No.112005(11)：72-75.
© 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 |