基于压缩感知的毫米波成像图像重构研究

发布时间：2019-05-05 09:02

【摘要】：毫米波成像可用于安检、飞机盲降、战场环境成像等领域,是当前的研究热点。目前,被动毫米波成像系统多采用多通道扫描体制实现快速成像,但存在通道一致调试较难以及成本较高等问题。本文尝试结合压缩感知理论,完成单通道被动毫米波编码成像系统,主要围绕编码和图像重构方法等开展研究,主要工作如下。(1)完成了毫米波单通道编码成像方案的设计。在分析被动毫米波成像机理的基础上,对单通道编码成像的光路进行了分析、计算和设计,结合压缩感知理论,建立了毫米波单通道编码成像数学分析模型。(2)完成了基于LDPC码校验矩阵的毫米波单通道成像编码模板设计。在分析比较Hadamard矩阵、循环S矩阵等几种典型测量矩阵基础上,结合实际工程实现的需求,提出采用基于PEG—准循环构造法的LDPC码校验矩阵作为编码模板的测量矩阵,仿真结果验证了其满足压缩感知测量矩阵的基本条件并相较其它方法具备重构性能优、结构性强、可循环性的优势,同时完成了实际基于LDPC码校验矩阵作为编码模板的硬件设计。(3)采用TV正则化的结构聚类超分辨图像重构算法实现单通道编码成像的图像重构。在深入研究多种基于压缩感知图像重构方法基础上,重点研究稀疏表示下毫米波图像的超分辨率重构,对比联合K-SVD字典的稀疏表示方法和结构类聚方法的重构性能,并在结构类聚超分辨方法基础上采用TV正则优化对算法进行优化,仿真结果验证了该方法的有效性。
[Abstract]:Millimeter wave imaging, which can be used in security inspection, aircraft blind landing, battlefield environment imaging and so on, is a hot research topic at present. At present, many passive millimeter wave imaging systems use multi-channel scanning system to realize fast imaging, but there are some problems such as the difficulty of channel uniform adjustment and the high cost of passive millimeter wave imaging system. This paper attempts to complete a single-channel passive millimeter-wave coded imaging system based on compression sensing theory, mainly focusing on coding and image reconstruction methods, and so on. The main work is as follows. (1) the design of millimeter wave single channel coding imaging scheme is completed. On the basis of analyzing the mechanism of passive millimeter wave imaging, the optical path of single channel coded imaging is analyzed, calculated and designed, combined with compression sensing theory. The mathematical analysis model of millimeter-wave single-channel coding is established. (2) the design of millimeter-wave single-channel imaging coding template based on LDPC code check matrix is completed. Based on the analysis and comparison of several typical measurement matrices, such as Hadamard matrix and cyclic S matrix, combined with the requirements of practical engineering implementation, the LDPC code check matrix based on PEG- quasi-cyclic construction method is proposed as the measurement matrix of coding template. The simulation results show that it satisfies the basic conditions of compressed sensing measurement matrix and has the advantages of better reconstruction performance, stronger structure and circularity than other methods. At the same time, the hardware design based on LDPC code check matrix is completed. (3) using TV regularization structure clustering super-resolution image reconstruction algorithm to realize the single-channel coded image reconstruction. On the basis of deep research on many kinds of compressed perceptual image reconstruction methods, the super-resolution reconstruction of millimeter-wave images based on sparse representation is studied in detail. The performance of the sparse representation method combined with K-SVD dictionary and the structure clustering method are compared. On the basis of the structure clustering super-resolution method, TV regular optimization is used to optimize the algorithm. The simulation results show that the proposed method is effective.
【学位授予单位】：南京理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41

【参考文献】