CASIA OpenIR  > 毕业生  > 硕士学位论文
自发荧光图像分割算法的研究
Alternative TitleStudies on Segmentation algorithm of Bioluminescent Images
常志军
Subtype工学硕士
Thesis Advisor田捷
2010-05-29
Degree Grantor中国科学院研究生院
Place of Conferral中国科学院自动化研究所
Degree Discipline计算机应用技术
Keyword自发荧光成像 图像处理 医学图像分割 直方图 自适应分割算法 Bioluminescence Imaging Image Processing Medical Image Segmentation Histogram Adaptive Segmentation Algorithm
Abstract光学分子影像技术中自发荧光成像(BLI)因其对活体小动物具有特异性分子探针、高灵敏性、非侵入性以及可连续性观测等特点而被认为是一种较有应用前景的分子影像成像模态。因此,在国内研制一套具有自主知识产权的自发荧光成像系统具有重要意义。本文的主要工作即为对自发荧光图像分割理论和自发荧光成像系统研发进行了系统的分析与设计。自发荧光图像分割是系统的核心环节,其分割结果直接决定着可视化效果,进而影响图像分析,劣质的分割最终导致应用的不准确性,甚至出现应用错误。通过分析自发荧光图像的特点,本文首先提出了基于一维直方图的自动分割算法。自动分割算法需要先进行全局像素值总和归一化处理,再根据先验知识对ROI进行分割,从而获得分析目标。针对自动分割算法无法区分多光源的弊端,本文在自动分割的基础上继而提出了自适应分割算法,通过种子点屏蔽干扰区域并完成区分光源的目的。针对二维直方图的分割算法能较好的处理荧光图像中坏点问题,本文最后又提出基于荧光图像分割的灰度-圆周域二维直方图的分割算法,为提高分割性能,使得前景与背景的整体区分度最大,采用类间减类内方差最大法的测度准则。本团队自主研发的自发荧光成像系统(WinMI)主要包含硬件数据采集,图像处理和图像分析三部分。在图像处理过程中,对自发荧光图像进行分割处理前需进行降低噪声的处理,分割后要进行加伪彩以提高可视化效果,并且为确定分割目标的位置需要对荧光图像与白光图像进行图像融合。为解决图像处理中各环节的串行结构,本文提出了链式设计的思路,实现了高内聚、无耦合的目标,方便、高效地把图像分割算法及其它处理在WinMI中进行添加或删除。实验表明,自发荧光图像分割算法的研究与WinMI平台的架构设计都较好的满足了应用需求,得到了理想的处理结果,达到了预期的目标。最后,该自发荧光成像系统现已顺利通过第三方质量认证,并且成功实现了技术成果转移。
Other AbstractBioluminescence Imaging (BLI) of Optical Molecular Imaging is considered to be a more promising molecular imaging mode because of specific molecular probes, highly sensitivity , non-invasive and continuous observations. So it is very significant to development a bioluminescence imaging system owned independent intellectual property rights in China. This paper provided analysis and design from image segmentation theory to a bioluminescence imaging system architecture. Bioluminescent images segmentation is a core part of BLI system, and segmentation results directly determines the visualization, thereby affecting the image analysis, resulting application inaccuracy even application error. This paper firstly analyzed the characteristics of bioluminescent images , then proposed both automatic segmentation algorithm, which based on one-dimensional histogram. The first step of automatic segmentation is to normalized every pixel by globel sum of pixels of the bioluminescent image, then segment the ROI with prior knowledge to obtain the target. This paper proposed adaptive segmentation to remedy the defects of Automatic segmentation algorithm for multi-source, which shields the interference region and distinguishes the resources by seed points. Segmentation algorithm based on tow- dimensional histogram of gray and circle mean value can better handle the “dead pixel” issue of bioluminescent images, and this paper used the measure criteria of maximizing the intra-class variance minus inter-class variance to improve the segmentation performance and achieve the overall distinction between foreground and background. Bioluminescence Imaging System (WinMI) mainly contains data collection, image processing and interactive image analysis. It is required to reduce the affect of noise before segmenting the bioluminescent images, and it is required too to add pseudo-color after segmenting the bioluminescent images and to fuse the white image and bioluminescent image to determine the position of ROI. To solve the problem of serial structure of image processing , this paper proposed chain design, realizing the goal of high cohesion and no coupling ,which could expediently and efficiently adds or deletes image segmentation and other processing procedure to WinMI. Experiments showed that both bioluminescent image segmentation algorithm and platform of bioluminescence imaging system can well meet the application requirements, achieved the expected goal. Finally, WinMI has pa...
shelfnumXWLW1562
Other Identifier200728017029251
Language中文
Document Type学位论文
Identifierhttp://ir.ia.ac.cn/handle/173211/7531
Collection毕业生_硕士学位论文
Recommended Citation
GB/T 7714
常志军. 自发荧光图像分割算法的研究[D]. 中国科学院自动化研究所. 中国科学院研究生院,2010.
Files in This Item:
File Name/Size DocType Version Access License
CASIA_20072801702925(4605KB) 暂不开放CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[常志军]'s Articles
Baidu academic
Similar articles in Baidu academic
[常志军]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[常志军]'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.