Non-invasive multispectral scattering imaging via OTF retrieval

Abstract

Spectral imaging can obtain spatial and spectral information of an object and play an important role in many application scenarios. Computational spectral imaging through scattering media utilizes the spectrally sensitive properties of scattering media as a filter for spectral imaging. The spatial and spectral information are reconstructed simultaneously by deconvolution with point spread function (PSF). Invasive spectral imaging imposes stringent constraints on the measurement accuracy of PSF and application scenarios. Here, we demonstrate a non-invasive multispectral scattering imaging method via optical transfer function (OTF) retrieval. The method uses multi-frame speckles to non-invasively retrieve the OTF of imaging system. Similar to the spectral filter property of PSF in the space domain, OTF at different wavelengths can be employed to filter and reconstruct the multispectral information of mixed speckle in the frequency domain. Our method overcomes the need for invasive measurements and is applicable to a wide range of scenarios for static and dynamic objects, providing a new approach to multispectral imaging.