中国科学院光电技术研究所机构知识库
Advanced  
IOE OpenIR  > 光电技术研究所博硕士论文  > 学位论文
题名:
基于扩展目标的跟踪与高精度定位技术
作者: 张建伟
学位类别: 博士
答辩日期: 2009-06-01
授予单位: 中国科学院光电技术研究所
授予地点: 光电技术研究所
导师: 张启衡
关键词: 目标跟踪 ; 扩展目标 ; 模板匹配 ; 局部特征 ; 亚像素 ; 仿射变换
其他题名: Tracking and High-precision Localization Technology Based on Extended Target
学位专业: 信号与信息处理
中文摘要: 目标跟踪技术是光电探测系统的关键技术之一,而扩展目标跟踪技术的研究对于目标跟踪技术的发展具有极其重要的意义。由于扩展目标在其视场内占有较大比例甚至目标已经超出视场范围,因此和典型的中小目标跟踪有所不同,在对扩展目标进行跟踪的时候,经典的目标跟踪方法已不再适用,因而,对扩展目标成像特征的深入研究,采取稳定的跟踪方法,进而得到高精度测量定位,就显得格外重要。本论文针对其技术难点,在理论和算法上进行了深入的分析和研究,提出了解决方法,并取得了阶段性进展。 对于扩展目标而言主要有以下特点:其一扩展目标在视场内占据比例较大,跟踪区域容易滑动。其二扩展目标占据视场较大,与小目标相比拥有更多地细节特征以及高频信息。其三扩展目标通常整体特征缺失但局部特征丰富,因此可以实现局部特征匹配。根据扩展目标的特性,论文着重围绕以局部特征匹配为基础的扩展目标稳定跟踪技术展开研究,在实现稳定跟踪的前提下,又研究了利用亚像素测量技术实现扩展目标精确定位的方法。 通过对扩展目标成像特性与目标跟踪技术基本方法的分析,确定了在以相关匹配法进行区域锁定的基础上利用局部特征匹配校正的扩展目标稳定跟踪模型。 针对普通人造刚体目标存在直线边缘的特点,提出了一种块遍历的直线边缘特征提取方法。该方法利用一个小窗口在整个图像上遍历,并通过自动阈值分割、Freeman链码、直线拟合等一系列操作检测出每个块内是否存在直线段,从而实现了一种检测直线边缘的方法。根据直线边缘的检测方法,将其应用到扩展目标跟踪中,由于直线交点有较好的可重复检测性,在MCD区域锁定的基础上,利用目标中的任意两条直线倾角与截距对直线交点建立描述子,并通过对相邻帧间的交点描述子的匹配,实现了扩展目标的稳定跟踪。实验结果表明,在扩展目标变形不大的情况下,该方法对扩展目标的跟踪有着较好的稳定性和鲁棒性。 为了拓展跟踪方法的适应性,利用基于尺度空间理论的SIFT(Scale Invariant Feature Transform)局部特征匹配来实现扩展目标的稳定跟踪:在MCD区域锁定的基础上,对图像建立Gaussian尺度空间并生成Difference of Gaussian 函数,进行位置与尺度的三维极值检测来得到稳定的SIFT局部特征,然后使用高斯尺度空间的梯度大小和方向来建立各个特征点的描述子,当相邻帧间匹配的特征点对达到三对时,就可以计算出目标变换的仿射参数,由此可以准确的计算出目标上初始跟踪点在现行帧中的位置以实现扩展目标的稳定跟踪。由于SIFT特征的对旋转、尺度甚至一定程度上对仿射具有不变性,因此保证了该方法可以实现扩展目标稳定跟踪。 结合扩展目标的特点对亚像素测量技术进行了系统的分析与研究,采用双线性插值对局部模型法进行改进以达到提高局部特征拟合精度的目的。然后将相关函数法与SSDA(Sequence Similarity Detection Algorithm)相结合实现扩展目标的精确定位。实验表明改进的局部模型法的特征拟合精度得到了明显的提高,相关函数拟合法与SSDA结合将定位精度精确到了0.1-0.2个像素。由此来看,亚像素测量技术对目标定位精度的提高有着非常重要的作用,作为一种节省硬件成本的方法拥有非常实际的意义。 结合工程需要,对扩展目标跟踪算法的实时性进行了检验,经过对SIFT扩展目标跟踪算法的进一步优化,完成了在基于TMS320C6455的图像信号处理板上的实验验证,并显示了良好的跟踪效果。 本文针对扩展目标的跟踪与高精度定位的难点,结合工程实际,提出了针对性较强的算法和方法,并在工程应用中取得了良好的效果,为提高光电成像跟踪系统对扩展目标的跟踪水平提供了有力的技术支持。
英文摘要: The target tracking has been one of the key techniques of optical-electro detecting system, the extended target tracking plays an important role in this field. Because extended target is almost full of the whole image even part of extended target is out of the image, the tracking methods of extended target are different from the ones of the typical target, the methods used in typical target tracking, is not suitable in the extended target tracking. So, deeply analyzing the characteristic of extended object, studying more stable tracking method and obtaining the higher accuracy are very significant. In this dissertation for Ph.D degree, elaborate analysis and study were performed and some effective algorithms were developed to solve some difficulties in extended target tracking, meanwhile some important progresses were made. The extended target mainly includes three characteristics: first, the object is large and the tracking position drifts off easily; second, extended object owns more details and high-frequency information; third, due to loss of the whole feature and enough local feature, local feature can be used to match. According to the characteristics, the focus of research is stable tracking based on the local feature matching in locked local region, and then the research on the accurate localization based on sub-pixel is done. The development, the relative fundamental theories and the methods about target tracking and local feature were discussed and researched, the extended object tracking model based on local feature matching was given. For obtaining a kind of stable local feature, an algorithm of extracting straight-line edge feature based on ergodic block is proposed. A small window moves through the whole image, while the window move each step, the image in the window will be processed by several steps including auto-gray threshold, Freeman chain-edge extraction, straight-line fitting and so on. Consequently the straight-line edge is detected. The method of straight-line detection is used in extended target tracking, due to good repeatability of intersections, the straight-line parameters-obliquities and intercepts are used to produce the descriptors of intersections which are formed by arbitrary two straight-lines, and then the intersections are matched between neighboring frames by computing the Euclidean distance of descriptors, finally the centroid of intersections is considered as the tracking position. Experimental results show that if only the target does not transform strongly, the tracking method is stable and robust. In order to extend the adaptability of extended target tracking method, SIFT feature, which is based on scale space theory, is used in extended target tracking. According to scale spaces theory, firstly Gaussian scale spaces are built, and then Difference of Gaussian (DoG) are formed, SIFT feature is detected in DoG and described by gradient magnitude and orientation of Gaussian scale spaces. If at least three pairs of SIFT feature points can be matched in neighboring frames, the affine parameters of target transformation between frames can be computed. By affine parameters the accurate position of initial tracking point in current frame is computed. Due to the invariant of SIFT feature to rotation, scale even to affine transformation to some extent, the method has a good performance in extended target tracking. According to the trait of extended target, the sub-pixel technique is analyzed and researched, bilinear interpolation is used to improve the local model method to obtain better feature fitting accuracy and then relative function method and SSDA is used to realize the more precise extended target localization. The experiment shows that above methods have good effects and the localization precision is up to 0.1-0.2 pixel. So, sub-pixel technique is very important for improving tracking precision as a method which saves the hardware cost. In order to satisfy the engineering requirement of the extended target tracking algorithm, SIFT tracking method is optimized and implemented in the image processing platform based on TMS320C6455, and experimental results show the tracking method has good performance. In conclusion, some innovated methods are developed to solve the difficulties in extended target tracking and high-precision localization and engineering practicality. The research works of this dissertation has important reference to the future development of the the extended target tracking in optical-electro detecting system.
语种: 中文
内容类型: 学位论文
URI标识: http://ir.ioe.ac.cn/handle/181551/348
Appears in Collections:光电技术研究所博硕士论文_学位论文

Files in This Item:
File Name/ File Size Content Type Version Access License
10001_200618015127023张建伟_word2003.doc(9174KB)----限制开放View 联系获取全文

Recommended Citation:
张建伟. 基于扩展目标的跟踪与高精度定位技术[D]. 光电技术研究所. 中国科学院光电技术研究所. 2009.
Service
Recommend this item
Sava as my favorate item
Show this item's statistics
Export Endnote File
Google Scholar
Similar articles in Google Scholar
[张建伟]'s Articles
CSDL cross search
Similar articles in CSDL Cross Search
[张建伟]‘s Articles
Related Copyright Policies
Null
Social Bookmarking
Add to CiteULike Add to Connotea Add to Del.icio.us Add to Digg Add to Reddit
文件名: 10001_200618015127023张建伟_word2003.doc
格式: Microsoft Word
此文件暂不支持浏览
所有评论 (0)
暂无评论
 
评注功能仅针对注册用户开放,请您登录
您对该条目有什么异议,请填写以下表单,管理员会尽快联系您。
内 容:
Email:  *
单位:
验证码:   刷新
您在IR的使用过程中有什么好的想法或者建议可以反馈给我们。
标 题:
 *
内 容:
Email:  *
验证码:   刷新

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

 

 

Valid XHTML 1.0!
Copyright © 2007-2016  中国科学院光电技术研究所 - Feedback
Powered by CSpace