压缩感知鬼成像中观测矩阵构造

发布时间：2018-05-02 13:39

本文选题：压缩感知(CS) + 鬼成像　；参考：《光电子·激光》2016年12期

【摘要】：为了提高压缩感知鬼成像(CSGI)的成像效果,解决现阶段观测矩阵存在的稳定性差、数据庞大和不满足非负性等问题,对观测矩阵的构造方法进行研究。首先,介绍确定性随机序列的产生方法和性质,可以用作观测矩阵,满足有限等距要求。针对光强的非负性,提出利用偶次幂的余弦函数产生确定性随机序列的方法,构造观测矩阵并证明其性质;然后,通过仿真验证该观测矩阵的正确性,研究了序列的初始值和函数的幂对矩阵重构性能的影响;最后,搭建实验平台,对比常用的高斯随机矩阵(GM),分析本文方法的适用性和优缺点。实验结果表明,在鬼成像中,利用本文所构造的随机矩阵,重构图像峰值信噪比(PSNR)与GM相当,但存储的数据量大大减少,可满足鬼成像系统的快速高效、简单方便和成本低等要求。
[Abstract]:In order to improve the imaging effect of compressed perceptual ghost imaging (CSGI) and to solve the problems of poor stability, huge data and non-negativity of the observation matrix at present, the method of constructing the observation matrix is studied. Firstly, the generation methods and properties of deterministic random sequences are introduced, which can be used as observation matrices to meet the requirements of finite equidistant. In view of the nonnegativity of light intensity, a method of generating deterministic random sequences by using the cosine function of even power is proposed, the observation matrix is constructed and its properties are proved, and the correctness of the observation matrix is verified by simulation. The effects of the initial value of the sequence and the power of the function on the reconstruction performance of the matrix are studied. Finally, an experimental platform is built to compare the common Gao Si random matrix with GM.The applicability, advantages and disadvantages of this method are analyzed. The experimental results show that the PSNRR of reconstructed image is equivalent to that of GM by using the random matrix constructed in this paper, but the amount of data stored is greatly reduced, which can satisfy the fast and high efficiency of ghost imaging system. Simple and convenient and low cost requirements.
【作者单位】：电子工程学院脉冲功率技术国家重点实验室;
【基金】：国家自然科学基金(61271376) 安徽省自然科学基金(1208085MF114)资助项目
【分类号】：TP391.41;O431.2

【相似文献】