中国科学院沈阳自动化研究所机构知识库
Advanced  
SIA OpenIR  > 机器人学研究室  > 学位论文
题名: 基于先验知识的图像分割与目标跟踪
其他题名: Image Segmentation and Object Tracking Based on Prior Knowledge
作者: 朱琳琳
导师: 唐延东
分类号: TP391.4
关键词: 主动轮廓 ; 先验知识 ; 图像分割 ; 目标跟踪
索取号: TP391.4/Z82/2010
学位专业: 模式识别与智能系统
学位类别: 博士
答辩日期: 2010-11-29
授予单位: 中国科学院沈阳自动化研究所
学位授予地点: 中国科学院沈阳自动化研究所
作者部门: 机器人学研究室
中文摘要: 图像分割与目标跟踪都是计算机视觉中的基本任务。图像分割是图像处理中的一项最重要的基本任务,也是图像分析的前提和基础。而基于视觉的目标跟踪可以获得目标的各种运动参数,从而进行进一步处理与分析,实现对运动目标的行为理解,以完成更高一级的任务。经过多年的发展,已经出现了很多种的方法来解决这两个问题。分割算法分为基于区域、基于边缘和基于特定理论的,还有一类分割算法利用序列图像间运动信息对图像进行分析。跟踪算法分为基于特征点的跟踪,基于区域的跟踪和基于轮廓的跟踪。这些算法各有特点,可以解决不同情况下的图像处理问题,但是它们在过程中还是不能充分利用高层先验知识影响底层的图像数据处理,从而无法模拟人类视觉系统智能性。 本文的一个主要研究方向是如何正确、合理地将目标的先验知识应用到基于主动轮廓的图像分割与目标跟踪算法中,这也是主动轮廓方法的发展趋势; 另外,如何将光学成像的先验知识加入到图像处理过程也是本文的重要研究内容。主要工作内容及成果有如下四部分: 1、基于目标形状特征的管状目标主动轮廓分割方法:在利用主动轮廓对管状目标进行分割时,期望主动轮廓可以沿着目标的延展方向生长。基于这一曲线演化的先验知识,首先在目标增强的基础上提取目标的延展方向,然后改变主动轮廓的演化方程,用延展方向控制曲线的演化方式,使得主动轮廓可以沿着管状目标演化,达到快速准确的管状目标分割。该算法可应用于医学图像中的血管分割和河流、道路等管状目标的分割。 2、基于梯度下降流的主动轮廓快速演化方法——两步法:当主动轮廓初始曲线远离被分割目标或者用于目标跟踪过程时,其演化过程一般具有先平移后变形的趋势,而曲线演化的状态与其平均梯度下降流有密切关系。利用这一先验知识,本文将主动轮廓的演化分为先全局平移后局部形变两个步骤,提出了具有独立的全局平移与局部形变的主动轮廓方法。该方法基于对H0内积空间中的梯度流分析,可以用来对目标进行高效的分割或跟踪。 3、基于目标先验分布的主动轮廓跟踪方法:将目标的分布作为先验知识加入主动轮廓的能量泛函中,提出了面向目标分布的C-V改进模型,提高主动轮廓跟 踪算法的准确性。同时利用Mean Shift的快速收敛性和计算产生的目标概率模型,提出了主动轮廓与Mean Shift混合跟踪算法,在保证算法鲁棒性的基础上提高主动轮廓跟踪算法的计算速度。 4、提出了具有阴影抵抗力的动态目标检测与跟踪方法:阴影给很多视觉问题带来了困难。彩色图像的三色衰减模型描述了阴影区域衰减值的衰减关系,以此衰减关系作为先验知识,去除动态目标检测时的目标投射阴影问题;而且在三色衰减模型的基础上提出了阴影不变特征,设计了阴影背景下的目标特征选择方式,提出了对阴影具有鲁棒性的自动目标跟踪算法。试验结果显示了该算法可有效消除阴影对目标跟踪的干扰,解决了部分室外阴影条件下的目标跟踪问题。
英文摘要: Image segmentation and object tracking are the basic tasks in computer vision. Image segmentation is one of the most important processing before all the image analysis. Vision-based target tracking can get all kinds of motion parameters for further processing and analysis in order to achieve understanding of the behavior of moving objects for higher level tasks. After years of development, there are many methods proposed to solve these two problems. Segmentation algorithm can be divided into region based, edge based and specific theory based. There is also a class of image segmentation algorithm make use of sequence information between the motion of the sequence images. Tracking algorithm is divided into feature point based tracking, region-based tracking and silhouette tracking. These algorithms have their own characteristics, and different methods can solve image processing problems in different conditions. However they still can’t take full advantage of high-level prior knowledge to the image data processing, and thus can not simulate the intelligence of the human visual system. The main research direction in this dissertation is how to use the prior knowledge rightly and effectively for image segmentation and target tracking. Using the prior knowledge of optical imaging to image processing is other important research of this dissertation. The main contents and contributions are summarized as follows. 1. An Active Contour Method for Tubular Object Segmentation. In segmenting the tubular object, we hope the contour will evolve along the object extending direction. Base on this prior knowledge, we get the tubular structure response intensity as well as the response direction, and the response result is incorporated in the active contour model to control the curve evolution. 2. Two-Step Active Contour Method Based on Gradient Flow (Active Contour Method with Separate Global Translation and Local Deformation). When a contour curve is far away from the object to be segmented or tracked, its evolving toward to the object can be resolved into two processes: global translation and local deformation. Based on prior knowledge, the contour evolution process is separated into two steps: global translation and local deformation. The contour global translation and local deformation are realized by average gradient flow and normal gradient flow of the evolving contour curve respectively. 3. Target’s prior distribution based active contour tracking. The feature distribution of target can be get before the tracking. By putting the prior distribution into the active contour’s energy functional, we propose a improved C-V active contour model for tracking. Meanwhile, we propose also a hybrid active contour tracking algorithm with the Mean Shift. It uses Mean Shift and the rapid convergence of the target computed probability model to improve the computational speed and robustness. 4. Shadow Resistant Tracking and detection via Tricolor Attenuation Model. Shadows, the common phenomena in most outdoor scenes, bring many problems in image processing. Tricolor attenuation model (TAM) presents the description of the relationship of the attenuation in each color channel. Based on this relationship, we propose a method to remove the object’s cast shadow of the target in detection. By using a shadow invariant transform, we propose a shadow resistant tracking method for eliminating the shadow disturbance to object tracking.
语种: 中文
产权排序: 1
内容类型: 学位论文
URI标识: http://ir.sia.cn/handle/173321/9389
Appears in Collections:机器人学研究室_学位论文

Files in This Item:
File Name/ File Size Content Type Version Access License
基于先验知识的图像分割与目标跟踪.pdf(4688KB)----限制开放 联系获取全文

Recommended Citation:
朱琳琳.基于先验知识的图像分割与目标跟踪.[博士 学位论文 ].中国科学院沈阳自动化研究所 .2010
Service
Recommend this item
Sava as my favorate item
Show this item's statistics
Export Endnote File
Google Scholar
Similar articles in Google Scholar
[朱琳琳]'s Articles
CSDL cross search
Similar articles in CSDL Cross Search
[朱琳琳]‘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
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  中国科学院沈阳自动化研究所 - Feedback
Powered by CSpace