为了改善传统的多幅亚像素图像配准融合实现超分辨率的方法面临的配准误差和高成本问题,将压缩传感理论引入超分辨率成像.基于大多数自然图像普遍具有的稀疏表示特性,以经典的4-f光学架构为基础,利用频域中相位比振幅包含更多信息的特点,提出了一种频域纯相位调制压缩成像方法,通过重建算法从单次曝光记录的低维测量值中恢复原高分辨率图像的信息.数值实验结果表明,提出的方法可以有效地实现图像信息的随机调制和高质量重建,是一种有潜力的压缩成像物理实现方案,具有较高的重建信噪比和较少的重建时间,尤其是对于大尺度图像.
One of the goals of optical imaging and image processing is super-resolution imaging. In order to reduce the registration error and costly problem facing in multiple sub-pixel image registration fusion method to achieve super-resolution, a compressive sensing method is introduced for super-resolution imaging, it benefits from the general sparse representation feature of most nature images. Based on the classical 4-f optical architecture, the phase will contain more information than the amplitude in frequency domain, a compressed imaging method with pure phase modulation in the frequency domain is proposed. The original high-resolution image information can be recovered from the low-dimensional measurements recorded with a single exposure by various algorithms. Numerical results demonstrate that the proposed can effectively achieve random modulation of image information and high-quality reconstruction, which can be considered as a promising scheme for physics implementation construction signal to noise ratio and less reconstruction time, especi of compressed imaging with high really for large-scale image.