SIA OpenIR  > 光电信息技术研究室
Alternative TitleResearch on Automatic Target Detection and Recognition in Infrared Imaging Guidance
Thesis Advisor于海斌 ; 史泽林
Keyword红外图像 Atr 小波变换 数学形态学 特征选择和提取
Call NumberTP391.4/W58/2004
Degree Discipline机械电子工程
Degree Name博士
Degree Grantor中国科学院沈阳自动化研究所
Place of Conferral沈阳
AbstractATR是红外成像制导的核心技术之一,本论文围绕红外图像ATR的技术难题进行深入的理论分析及实验研究。主要完成了以下工作: 1.设计了行均值相减、小波去噪和自适应维纳滤波,与形态Top-hat滤波相结合的预处理方法,有效去除红外图像的混合噪声,抑制复杂景物的结构性干扰。 2.提出基于小波多分辨率分析的复杂海空背景红外小目标检测算法。分别利用小波分解和Otsu法进行海天线检测;提出基于小波变换的水平垂直边缘交叉及多方向细节融合的目标检测算法。实验表明,上述算法具有良好的适应性和较高的目标检测率。 3.提出基于数学形态学的复杂背景条件下斑块状红外小目标检测算法。实验表明,该算法具有良好的适应性和稳健性,能够有效检测出海面、地面和公路背景低信噪比红外图像中的机动小目标。 4.提出直接采用小波多尺度自适应阈值,结合形态滤波处理,提取完整目标边界的模糊图像分割新思路。实验表明,该方法对多种模糊图像,具有较高的分割精度,对噪声和目标大小变化不敏感。5.利用Hu不变矩构造了一组新的综合矩不变量IMI,在Hu矩的RTS不变性证明的基础上,证明了IMI对二维离散图像具有RTS不变性,同时具有对比度和照度不变性,而且对灰度、区域或边界图像具有统一的计算形式。
Other AbstractAutomatic Target Recognition (ATR) is a key technology in infrared (IR) imaging guidance. This dissertation researches into some pivotal and important problems in the chain of ATR through theoretic analysis and the experiments. Main contributions of the dissertation are summarized as follows: 1. In this thesis, image enhancement algorithms, such as the row gray average subtraction, wavelet-based denoising approach and adaptive Weiner filtering combining the morphologic Top-hat filtering are designed to improve the contrast, remove the mixed noises, correct blur and suppress the structural clutters in IR images. 2. The automatic detection algorithms based on wavelet multi-resolution analysis are presented for small targets in infrared images taken in complicated sea-sky background. Two SSDL detection algorithms based on wavelet decomposition and Otsu method are proposed in this thesis. Then, two small IR target detection algorithms are presented. One algorithm adopts the intersection of horizontal and vertical edges of the proper Approximation and another algorithm fuses the various direction Details to suppress noise and catch the candidate targets in the TPR. Experiments show these algorithms are adaptive and with high target detection ratio. 3. Several algorithms are developed for spot IR target detection based on the mathematical morphology. Many experiments indicate that these algorithms are adaptive, efficient, robust and accurate for detection and segmentation the small mobile targets in low SNR infrared images taken in complicated background of sea, land and road. 4. A new idea for segment and recognize the blurred targets in IR images. An adaptive threshold method are proposed on the basis of the Wavelet Multi-Resolution Filtering (WMRF), combined morphological filter to gain the whole boundary of target in this thesis. Experiments show this method is with high segmentation accuracy and low sensitivity to noise and scale change of targets. 5. The limitation of Hu moment and some moment in existence is analyzed theoretically. Consequently, Hu moments are assembled to derive some new moment invariants, called Integrative Moment Invariants (IMI), whose invariability are integrated in all cases, e.g., discrete image affine transform (Rotation, Translation, Scaling, RTS), change of contrast, noise and blur, shape region or boundary.
Contribution Rank1
Document Type学位论文
Recommended Citation
GB/T 7714
温佩芝. 红外成像制导自动目标检测与识别方法研究[D]. 沈阳. 中国科学院沈阳自动化研究所,2004.
Files in This Item:
File Name/Size DocType Version Access License
红外成像制导自动目标检测与识别方法研究.(3972KB) 开放获取LicenseApplication Full Text
Related Services
Recommend this item
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[温佩芝]'s Articles
Baidu academic
Similar articles in Baidu academic
[温佩芝]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[温佩芝]'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.