Optical Calibration of a Digital Micromirror Device (DMD)-based Compressive Imaging (CI) System

Compressive imaging (CI) system is a novel electro-optical imaging system design, which uses a single-pixel photo detector to capture two dimensional (2D) images. Instead of sampling the image directly by the sensor following the classic Nyquist-Shannon sampling theorem, CI systems insert a measurement layer between the image formation and the image recording media so that projection measurement matrices used to conduct compressive sampling can be effectively introduced to the imaging process. The Digital Micromirror Device (DMD) can be used to implement the projection measurement matrices. The imaging performance of a DMD based CI system relies more than just on the imaging optics and the pixel size of the sensor. It also depends on the design of the measurement matrices and their physical representations by the DMD. In the present work, we implemented three compressive sampling methods with the DMD, namely the random basis under-sampling method, the random sampling method in the Hadamard space and the variable density sampling method in the Hadamard space. We experimentally demonstrated that the design and implementation of these methods have a direct impact on the imaging performance of the CI system. We tested the system with different sampling ratios, DMD mirror configurations and imaging optics. Their influences on the reconstructed image quality are demonstrated by experimental results. Lastly, we discussed the illumination issue of the reconstructed image, which is not related to resolution, but is important for our visual perception of the reconstructed image.