摘要
本文根据双目CCD测距的研究现状及特点,对本课题进行了深入研究。通过对双目视差原理的分析,对CCD摄像机、镜头、测距平台等设备参数提出了具体要求,以满足测量远距离目标要求及高精度指标。从而最终确定并完成了双目CCD测距系统的总体设计方案。
由于摄像机标定是实现高精度的双目测量的前提条件,本文完成了对双目摄像机的标定。对Tsai的标定算法提出了改进,提出了一种完全线性标定摄像机的方法。根据径向平行约束和一阶径向畸变模型线性求解其余摄像机的内外部参数,包括焦距、一阶径向畸变系数、旋转矩阵和平移向量。实验中针对标定遇到的问题,改进了原有算法,采用单独标定平移分量tz,消除了tz对标定结果造成的不稳定影响,得到了收敛性较好的焦距值。实验数据表明该算法具有较高的标定精度。
本文还实现了高精度亚像素匹配,匹配算法是高精度双目测距的核心。采用匹配精度较高的互相关匹配算法,为了克服该算法运算较复杂的缺点,提出了一种基于尺度变换的快速匹配算法,大大节省了匹配所消耗的时间。根据互相关匹配结果获得的相关函数,对相关峰进行了分析,提出了两种拟合算法来进一步提高匹配精度,达到亚像素级别。
根据测量结果表明,本文的双目CCD测距系统具有较高的测量精度,目前可以测量3km以内的目标,实际测量精度高于0.5%。
According to the status and characteristics of binocular CCD cameras ranging, the subject has been studied depth in this paper. Through analyzing the binocular parallax, the specific requirements of the CCD camera, lens, ranging platform, and other equipments'parameters are proposed for satisfying the requirement of distant range target measurement and high-precision. The binocular CCD camera ranging system has been determined and completed in this paper.
Calibrating the camera is the precondition of high-precision binocular measurements, and the camera is calibrated perfectly in experiment. A completely linear approach for camera calibration is presented in this paper, and it improves the Tsai's calibration algorithm. Solving the other internal and external camera parameters based on Radial-Array-Constraints and first-order radial distortion model, which include focal length, first-order radial distortion parameter, rotation matrix and translation vector. Because of the problem in calibration experiments, the original algorithm has been improved. It calibrates the translation component tz, which eliminates the unstable effect for the result and makes the good convergence of the focal length. Experimental data show that the algorithm has high accuracy.
The article also achieved high precision sub-pixel matching, and the matching algorithm is the core of high accuracy binocular ranging. By applying the correlative matching algorithm, the fast matching algorithm based on scale transform is presented for overcoming the shortcomings of more complex operations, and it significant saves in time. According to the correlative function obtained in matching, two fitting algorithm are proposed for improving the matching accuracy, so as to achieving sub-pixel level.
Measurement results show that the binocular CCD camera ranging system has a high accuracy, and the measurement accuracy better than 0.28% within 3 km distance in the current.
引文
[1]马颂德,张正友.计算机视觉[M]：计算机理论与算法基础[M].北京：科学出版社,1998.
[2]祝琨.基于双目视觉的运动物体实时跟踪与测距：[硕士论文].北京：北京交通大学信号与信息处理系,2008.
[3]张广军.机器视觉[M].北京：科学技术出版社,2005.
[4]张虎.机器视觉中二维图像的三维重建：[硕士论文].北京：北方工业大学机械电子工程系,2006.
[5]刘鹏鑫,王扬.反求工程中多传感组合测量系统研究现状与发展[J]I机械科学与技术,2008,27(7)：896-901.
[6]童新华.双目视觉三维测量技术在地形测量中的应用：[硕士论文].北京：中国石油大学机械设计及理论系,2008.
[7]杜歆.用于导航的立体视觉系统：[博士论文].浙江：浙江大学通信与与信息系统系,2003.
[8]陈佳.CCD测距技术研究：[硕士论文].吉林：长春理工大学测试计量技术及仪器,2009.
[9]J. Mikko Hakkarainen. A 40×40 CCD/CMOS Absolute-Value-of-Difference Processor for Use in a Stereo Vision System[J]. IEEE JOURNAL OF SOLID-STATE CIRCUITS,1993:799～807.
[10]Steve Vallerand, Masayuki Kanbara, Naokazu Yokoya. Binocular vision-based augumented reality system with an increased registration depth using dynamic correction of feature positions[J]. Proceedings of the IEEE Virtual Reality,2003:271～272.
[11]李雪娟.基于双目视差的数字式测距望远镜：[硕士论文].天津：天津大学工程光学系,2007.
[12]樊学武,马臻,陈荣利,李英才.红外双波段卡塞格林光学系统设计[J].光子学报,2003,32(4)：463～465.
[13]J. Weng, P. Cohen and M. Herniou. Calibration of stereo cameras using a non-linear distortion model[J]. Proc. International Conference on Pattern Recognition,1990,246～253.
[14]天原嫄,张云辉,谭庆昌.CCD摄像机标定的研究[J].微计算机信息,2008,24(5)：206～208.
[15]Y I Abdel-Aziz, H M Karara. Direct linear transformation from comparator coordinates into object space coordinates in close-range photogrammetry[C]. In:Proceedings of the ASP Symposium on Close-Range Photogrammetry, Urbana, Illinois,1971:1～18.
[16]吴福朝,阮宗才等.非线性模型下的摄像机自标定[J].计算机学报,2002,25(3).
[17]Roger Y. TSAI. A versatile camera calibration technique for high-accuracy 3D machine vision metrology using off-the-shelf TV cameras and lenses[J]. IEEE Journal of Robotics and Automation,1987,3(4):323～334.
[18]Zheng-you ZHANG. A flexible new technique for camera calibration[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2000,22(11):1330～1334.
[19]冯翔.复杂三维模型相似性研究：[硕士论文].北京：北京师范大学计算机应用技术,2009.
[20]邹凤娇,苏显渝,李美菊.基于共面点的摄像机线性标定法[J].光电工程,2006,32(4).
[21]刘阳,王福利,常玉清,吕哲.黑白棋盘格角点检测算法[J].东北大学学报(自然科学版),2007,28(8)：1090-1093.
[22]谭晓军,郭志豪,蒋芝.基于对称性的棋盘方格角点自动检测算法[J].计算机应用,2008,28(6)：1540-1542.
[23]Lowe, D. G. Object Recognition from Local Scale-Invariant Features[C]. In Process of the International Conference on Computer Vision, pages 1150-1157, Corfu, Greece, September 1999.
[24]Lowe, D. G. Distinctive image features from scale-invariant key-points[J]. International Journal of Computer Vision,2004,60(2):91 ～ 110.
[25]Lenz Reimar K, Tsai Ronger Y. Techniques for calibration of the scale factor and image center for high accuracy 3D machine vision metrology[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,1988,10(5): 713～720.
[26]Lenz Reimar K, Tsai Ronger Y. Techniques for calibration of the scale factor and image center for high accuracy 3D machine vision metrology[J]. IEEE Transaction on Pattern Analysis and Machine Intelligence,1988,10(5): 713-720.
[27]胡海峰,熊银根.一种基于Hopfield网络的立体匹配方法[J].中国图形图像学报,2004,9(6)：729～736.
[28]Lloyd S A, etal. A paraller binocular stereo algorithm utilized dynamic programming and relaxation labelling[J]. CVGIP,1987,39:202～225.
[29]Aronld R D. Automated stereo perception[R]. Standard University Tech. Rep, STAN-CS-83-961,1983.
[30]Baker H H, Binford T O. Depth from edge and intensity-based stereo[D].7th Int. Joint Conf. Art. Intel, Vancouver,1981:631～636.
[31]谢海宴.立体视觉中的特征点匹配算法研究[J].宿州学院学报,2006,21(1)：96-98.
[32]华顺刚,曾令宜.一种基于角点检测的图像密集匹配算法[J].计算机工程与设计,2007,28(5)：1092-1095.
[33]胡明吴,任I明武,杨静宇.一种快速使用的特征点匹配算法[J].计算机工程,2004,30(9)：31-33.
[34]Barnard S, et al. Disparity analysis of images[J]. IEEE Trans. On PAMI,1980,2:333～340.
[35]宋长波.基于线特征的立体视觉匹配研究：[硕士论文].湖北：武汉大学测试计量技术及仪器,2004.
[36]A. Luo, W. Tao, H. Burkhardt. A new multilevel line-based stereo vision algorithm based on fuzzy techniques[C]. Proceeding of Int. Conf. on Pattern Recognition,1996, A:383～387.
[37]Ho. W. P., Yip, R. K. K. A dynamic programming approach for stereo line matching with structural information[D]. Proceedings of the 13th International Conference on Pattern Recognition,1996, A:791～794.
[38]Medioni G, etal. Segment-based stereo matching[J]. CVGIP,1985,31(1):2～18.
[39]李秀智,张广军.一种基于边缘线的三目立体匹配方法[J].光电工程,2007,34(2)：22-26.
[40]Ohta Y, et al. Stereo By intra-and inter-scan line search using dynamic programming[J]. IEEE Trans. On PAMI, 1987,7:139～154.
[41]Lee H H, etal. Region matching and depth finding for 3d objects in stereo aerial photographs[J]. Pattern Recognition,1990,23(2):81-93.
[42]刘正东,孙权森,杨静宇.基于特征约束及区域相关的体视匹配方法[J].计算机工程与应用,2003,3(4)：38～40.
[43]谢涛.基于区域分割的立体匹配及其相关研究：[硕士论文].辽宁：大连理工大学计算机应用技术,2005.
[44]孟晶晶.基于区域增长的立体匹配算法的研究：[硕士论文].辽宁：大连理工大学计算机应用技术,2006.
[45]刘畅,向键勇,杨斌利.一种新的基于梯度特征的快速相关匹配算法[J].激光与红外,2004,34(4)：305-307.
[46]Boaz J. Super, William N. Klarquist. Patch-based stereo in a general binocular viewing geometry[J]. IEEE Transactions On Pattern Analysis And Machine Intelligence,1997,19(3):247～252.
[47]雷鸣,张广军.基于互相关的图像匹配亚像素定位[J].光电工程,2008,35(5)：108-113.
[48]纪小刚,龚光容.准均匀三次B样条曲线的任意分辨率光顺.机械科学与技术,2009,28(11)：1478-1481.
[49]Yong Lv, Qibo Feng, Liangyu Qi. Sub-pixel surface fitting algorithm in digital speckle correlation method. The ninth international conference on electronic measurement&instruments,2009,4,959～962.
[50]潘岩.基于双目立体视觉的人脸三维重建方法的研究：[硕士论文].辽宁：大连理工大学机械制造及自动化,2005.
[51]祝世平,强锡富.基于坐标测量机的双目视觉测距误差分析[J].电子测量与仪器学报,2000,14(3)：26-31.
[52]陈胜勇,刘盛等.基于OpenCV的计算机视觉技术实现[M].北京：科学出版社,2008.
[53]刘佳音,王忠立,贾云得.一种双目立体视觉系统的误差分析方法[J].光学技术,2003,29(3)：354～360.
