稀疏采样傅里叶望远镜成像

发布时间：2018-03-09 15:15

本文选题：傅里叶望远镜　切入点：稀疏采样　出处：《光学精密工程》2010年03期 　论文类型：期刊论文

【摘要】：为了实现傅里叶望远镜快速成像,提出了一种稀疏采样图像重构方法,并对利用稀疏傅里叶样本精确重构目标图像的问题进行了研究。首先,基于压缩感知理论,并考虑目标图像在变换域的稀疏性或可压缩性,建立稀疏采样图像重构问题的优化模型。然后,构造适当的随机稀疏采样模板,对目标图像的傅里叶分量进行采样测量。最后,利用随机稀疏测量样本,通过非线性优化精确重构目标图像。实验结果显示,对实际的卫星图像,利用20%～30%随机测量样本非线性重构图像与利用全部测量样本直接重构图像的均方误差仅为4%～6%,表明利用随机稀疏傅里叶样本能够实现精确的图像重构,而且大大减少了测量样本的数量,从而有效降低了实现快速成像对傅里叶望远镜系统的成本和复杂性要求。
[Abstract]:In order to realize the fast imaging of Fourier telescope, a sparse sampling image reconstruction method is proposed, and the problem of using sparse Fourier sample to reconstruct target image accurately is studied. Considering the sparsity or compressibility of the target image in the transform domain, the optimal model of the sparse sampling image reconstruction problem is established. Then, the appropriate random sparse sampling template is constructed to sample and measure the Fourier component of the target image. Using random sparse measurement samples, the target image is accurately reconstructed by nonlinear optimization. The experimental results show that, for the actual satellite image, The mean square error between the nonlinear reconstruction of images using 20% random measurement samples and the image directly reconstructed with all measurement samples is only 4 / 6, which shows that accurate image reconstruction can be achieved by using random sparse Fourier samples. Moreover, the number of measurement samples is greatly reduced, which effectively reduces the cost and complexity of the Fourier telescope system to realize fast imaging.
【作者单位】：中国科学院长春光学精密机械与物理研究所;
【基金】：国家863高技术研究发展计划资助项目(No.2008AA7034250)
【分类号】：P111.5

【参考文献】