水下双目视觉系统中的目标分割和目标定位
|
摘要
针对典型灰度化和阈值方法用于水下目标分割时存在的目标分割不完整等问题,提出了基于灰度对比度增强的两步目标分割方法:第一步基于彩色矢量阈值进行目标粗分割;第二步分析局部对比度和区域内均匀性,动态确定不同通道的权值和反向处理,进而增强灰度对比度,完成目标精分割.针对典型标定、立体匹配和目标位置测量方法不适用于本研究环境的问题,提出了基于目标尺寸估计的目标定位方法,首先完成在线外参标定和目标圆心匹配,然后利用双目位置测量数据估计目标的尺寸,最后采用单目位置测量得到定位数据.结果表明:该分割方法能够准确分割目标,分割准确率和错分类比率均优于典型方法,定位方法能够得到目标三维位置数据,研制的双目视觉系统能够有效配合水下运载器-机械手系统完成自主作业任务.
Aiming at the segmentation problem of target partial segmentation by applying typical color-to-gray and thresholding methods,a two-step target segmentation method based on grayscale contrast enhancement was proposed.First,a rough target segmentation via color vector threshold was conducted.Second,the regional uniformity and the local contrast were analyzed to determine dynamically the weights of different channel and image reverse processing for enhancing the grayscale contrast,so the thresholding methods can complete accurate target segmentation.Aiming at the positioning problem that typical calibration,stereo matching and target position measurement methods are not suitable for the research environment,a target positioning method based on target-size estimation was proposed in this paper.First an online extrinsic parameter calibration and the center point of the target matching were conducted.Then the target-size was estimated by using a binocular position data.Finally the position data was obtained by adopting a monocular position measurement method.Underwater experimental results demonstrate that the proposed segmentation method can segment the target accurately,the segmentation accuracy(SA)and misclassification ratio(MCR)are all superior to typical methods,the proposed positioning method can obtain target 3 Dposition data,moreover,the developed binocular vision system can cooperate underwater vehicle-manipulator system(UVMS)for completing autonomous operations.
Aiming at the segmentation problem of target partial segmentation by applying typical color-to-gray and thresholding methods,a two-step target segmentation method based on grayscale contrast enhancement was proposed.First,a rough target segmentation via color vector threshold was conducted.Second,the regional uniformity and the local contrast were analyzed to determine dynamically the weights of different channel and image reverse processing for enhancing the grayscale contrast,so the thresholding methods can complete accurate target segmentation.Aiming at the positioning problem that typical calibration,stereo matching and target position measurement methods are not suitable for the research environment,a target positioning method based on target-size estimation was proposed in this paper.First an online extrinsic parameter calibration and the center point of the target matching were conducted.Then the target-size was estimated by using a binocular position data.Finally the position data was obtained by adopting a monocular position measurement method.Underwater experimental results demonstrate that the proposed segmentation method can segment the target accurately,the segmentation accuracy(SA)and misclassification ratio(MCR)are all superior to typical methods,the proposed positioning method can obtain target 3 Dposition data,moreover,the developed binocular vision system can cooperate underwater vehicle-manipulator system(UVMS)for completing autonomous operations.
引文
[1]Ji D H,Kim D,Kim J Y,et al.Redundancy analysis and motion control using ZMP equation for underwater vehicle-manipulator systems[C]∥IEEE Oceans.New York:IEEE,2016:1-6.
[2]李冀永,万磊,黄海,等.SY-Ⅱ水下机器人-机械手系统的协调运动控制[J].华中科技大学学报:自然科学版,2017,45(5):77-83.
[3]Koschan A,Abidi M A.Digital color image processing[M].New York:Wiley-InterScience,2008.
[4]Seo J W,Kim S D.Novel PCA-based color-to-gray image conversion[C]∥The 20th IEEE International Conference on Image Processing.New York:IEEE,2013:2279-2283.
[5]Otsu N.A threshold selection method from gray-level histogram[J].IEEE Transactions on Systems,Man,and Cybernetics,1979,9(1):62-66.
[6]Cai H M,Yang Z,Cao X H,et al.A new iterative triclass thresholding technique in image segmentation[J].IEEE Transactions on Image Processing,2014,23(3):1038-1046.
[7]徐德,谭民,李原.机器人视觉测量与控制[M].北京:国防工业出版社,2011.
[8]杨景豪,刘巍,刘阳,等.双目立体视觉测量系统的标定[J].光学精密工程,2016,24(2):300-308.
[9]桑瑞娟,王姮,张华,等.一种改进的区域双目立体匹配方法[J].传感器与微系统,2012,31(8):57-63.
[10]耿英楠.立体匹配技术的研究[D].长春:吉林大学图书馆,2014.
[11]Parmar J M,Patil S A.Performance evaluation and comparison of modified denoising method and the local adaptive wavelet image denoising method[C]∥Proc of International Conference on Intelligent Systems and Signal Processing.New York:IEEE,2013:101-105.
[12]Kim E,Haseyama M,Kitajima H.The extraction of circles from arcs represented by extended digital lines[J].IEICE Transactions on Information and Systems,2005,88(2):252-267.
[13]刘怀,黄建新.彩色图像的矢量阈值自适应分割算法[J].南京师范大学学报:工程技术版,2006,6(2):18-22.
[14]Bhujle H V,Chaudhuri S.Laplacian based non-local means denoising of MR images with Rician noise[J]Magnetic Resonance Imaging,2013,31(9):1599-1610.
[15]张铭钧,万媛媛,李煊.水中光视觉图像分割及目标提取方法[J].哈尔滨工程大学学报,2013,34(12):1580-1586.
[16]Levine M D,Nazif A M.Dynamic measurement of computer generated image segmentations[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,1985,7(2):155-164.
[17]Li C M,Huang R,Ding Z H,et al.A level set method for image segmentation in the presence of intensity inhomogeneities with application to MRI[J].IEEE Transactions on Image Processing,2011,20(7):2007-2016.
[18]Zhang Y,Brady M,Smith S.Segmentation of brain MR images through a hidden Markov random field model and the expectation maximization algorithm[J].IEEE Transactions on Medical Imaging,2001,20(1):45-57.
[2]李冀永,万磊,黄海,等.SY-Ⅱ水下机器人-机械手系统的协调运动控制[J].华中科技大学学报:自然科学版,2017,45(5):77-83.
[3]Koschan A,Abidi M A.Digital color image processing[M].New York:Wiley-InterScience,2008.
[4]Seo J W,Kim S D.Novel PCA-based color-to-gray image conversion[C]∥The 20th IEEE International Conference on Image Processing.New York:IEEE,2013:2279-2283.
[5]Otsu N.A threshold selection method from gray-level histogram[J].IEEE Transactions on Systems,Man,and Cybernetics,1979,9(1):62-66.
[6]Cai H M,Yang Z,Cao X H,et al.A new iterative triclass thresholding technique in image segmentation[J].IEEE Transactions on Image Processing,2014,23(3):1038-1046.
[7]徐德,谭民,李原.机器人视觉测量与控制[M].北京:国防工业出版社,2011.
[8]杨景豪,刘巍,刘阳,等.双目立体视觉测量系统的标定[J].光学精密工程,2016,24(2):300-308.
[9]桑瑞娟,王姮,张华,等.一种改进的区域双目立体匹配方法[J].传感器与微系统,2012,31(8):57-63.
[10]耿英楠.立体匹配技术的研究[D].长春:吉林大学图书馆,2014.
[11]Parmar J M,Patil S A.Performance evaluation and comparison of modified denoising method and the local adaptive wavelet image denoising method[C]∥Proc of International Conference on Intelligent Systems and Signal Processing.New York:IEEE,2013:101-105.
[12]Kim E,Haseyama M,Kitajima H.The extraction of circles from arcs represented by extended digital lines[J].IEICE Transactions on Information and Systems,2005,88(2):252-267.
[13]刘怀,黄建新.彩色图像的矢量阈值自适应分割算法[J].南京师范大学学报:工程技术版,2006,6(2):18-22.
[14]Bhujle H V,Chaudhuri S.Laplacian based non-local means denoising of MR images with Rician noise[J]Magnetic Resonance Imaging,2013,31(9):1599-1610.
[15]张铭钧,万媛媛,李煊.水中光视觉图像分割及目标提取方法[J].哈尔滨工程大学学报,2013,34(12):1580-1586.
[16]Levine M D,Nazif A M.Dynamic measurement of computer generated image segmentations[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,1985,7(2):155-164.
[17]Li C M,Huang R,Ding Z H,et al.A level set method for image segmentation in the presence of intensity inhomogeneities with application to MRI[J].IEEE Transactions on Image Processing,2011,20(7):2007-2016.
[18]Zhang Y,Brady M,Smith S.Segmentation of brain MR images through a hidden Markov random field model and the expectation maximization algorithm[J].IEEE Transactions on Medical Imaging,2001,20(1):45-57.