Computational Diffractive Sensing and Imaging: Using Optics for Computing and Computing for Optics

摘要

Computational diffractive imaging eschews the use of traditional lenses and curved mirrors and relies instead upon ultra-miniature diffractive optical elements paired with application-specific computational algorithms to achieve a desired end-to-end imaging function. The diffraction grating does not yield a traditional, human-interpretable image, but instead acts to preserve spatial information sufficient for estimation algorithms such as Tikhonov regularization or fast Fourier deconvolution to compute an image. Special purpose gratings can extract application-specific visual information for tasks such as motion estimation, point tracking, bar-code reading, and more.