IOE OpenIR  > 光电技术研究所博硕士论文
基于立体视觉水下电视测量方法的研究
陈志波
Subtype硕士
Thesis Advisor李声
2008-06-06
Degree Grantor中国科学院光电技术研究所
Place of Conferral光电技术研究所
Degree Discipline机械制造及其自动化
Keyword立体视觉 立体匹配 边缘提取 图像预处理 尺寸测量
Abstract本文提出了水下物体的3D测量方法,研究的主要目的是利用平行立体视觉对工业器件尺寸的测量。研究中的三个主要问题:摄像机的标定、水下环境中的图像预处理和立体图像对的匹配在本文中都得到了解决。 摄像机的标定主要是找出摄像机的内参数和摄像机系统本身的像差因子,所以它在3维测量系统中是一个非常重要的技术环节。在这个部分对立体视觉中的各种摄像机标定方法进行总结分类,重点介绍基于三维标定模板的线性标定方法、基于径向约束的Tai’s标定方法以及张氏平面标定法,简要地介绍了摄像机的自标定方法,并采用张氏平面标定法对摄像机标定进行仿真试验,给出了试验结果。 由于水下环境的特殊性质,在本研究中对于水下图像的预处理也是十分重要的,它对于后续的立体匹配和测量有着不可或缺的作用。其中图像的直方图均衡化方法用于增强图像的质量,中值滤波方法用于除去图像的噪声,这两个环节会使立体匹配变得更加稳健。由于要对器件进行尺寸测量,本文也介绍了几种边缘检测和图像分割的方法。 在立体视觉系统中,立体图像的匹配有着非常重要的作用。在第4章简要地介绍了一种基于特征的立体匹配方法,提出了采用构造特征链的方法使得匹配更具鲁棒性。第5章主要介绍了基于区域的匹配方法,这类方法可以分为两类:局部匹配方法和全局匹配方法。在局部方法中,对SSD、SAD和NCC几种方法进行了介绍,同时对处理结果进行了对比。本文研究了两种全局方法,提出了一种改进基于动态规划的匹配方法,另外实现一种是基于图割的方法,为了测量的准确性,本文采用基于图割的方法。
Other AbstractThis study presents a three-dimensional computer measurement system for underwater object.The main objective of this proposed research is to semi-automatically measure an industrial part in three-dimensions by using parallel stereo vision. Three major problems, camera calibration,image preprocessing for underwater enviroment and stereo matching, are solved. The camera calibration which is to maitain the camera interior parameters and camera’s distortion factor is the key technique for the 3D measurement system.We summarize and classify various methods of camera calibration of three-dimensional vision, mainly introduce the linearity calibration method based on 3D calibration template, the Tsai’s calibration method based on RAC and Zhengyou Zhang's flexible new technique for cameracalibration. Finally we carry on the emulation test of camera calibration using Zhengyou Zhang's flexible new technique and provide the result of the test. Because of the character of water, the image preprocessing is very important in this study. The histogram equalization technique was adopted to enhace the image quality and median filtering method to remove noise of image taken underwater. These two technique made the stereo matching processing more robust. For product dimension measurement, the edge detection and image segement methods were also introduced. The stereo matching is a significant problem in using a stereo vision system. In Chapter 4 we describe feature-based matching method which find corresponding features by feature chain. Chapter 5 is mainly about area-based stereo-matching meathod.These mothod can be classified into two class:local method and global method. For local method part, sum of squared differences, sum of absolute differences and normalized cross-correlation method were introduced and compared. Two global method, dynamic programming and graph cut, were studied in this research, and for measurement accuracy, graph cut method was adopted
Pages79
Language中文
Document Type学位论文
Identifierhttp://ir.ioe.ac.cn/handle/181551/304
Collection光电技术研究所博硕士论文
Recommended Citation
GB/T 7714
陈志波. 基于立体视觉水下电视测量方法的研究[D]. 光电技术研究所. 中国科学院光电技术研究所,2008.
Files in This Item:
File Name/Size DocType Version Access License
10001_20052801511900(18454KB) 开放获取CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[陈志波]'s Articles
Baidu academic
Similar articles in Baidu academic
[陈志波]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[陈志波]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.