Integral imaging (II), based on Integral Photography, is a promising technique for 3D object sensing, recording, and autostereoscopic visualization. Using this technique, a 3D scene can effectively be reconstructed by recording two-dimensional (2D) Elemental Images of the same 3D scene from different perspectives. II is therefore analogous to an incoherent version of conventional holography. Further, II systems can be used to reproduce detailed versions of 3D scenes that are comparatively better than those produced using conventional stereoscopic-based imaging techniques. Importantly, II is much easier to perform than classical hologram-based 3D imaging.

In some applications, the availability of low scattered illumination levels can lead to small irradiances at the image plane. In such cases, the existence of low-light-level or photon-counted imaging conditions must be considered. There are numerous techniques for estimating the equivalent normal intensity irradiance image from the counts of photon detectors in two-dimensional cases. The approaches are based on the statistical model of light fluctuation measurements, which declares that the photon count statistics follow a Poisson distribution. Therefore, the problem of reconstructing photon-counted images is equivalent to solving a Poisson inverse problem.

The majority of optical image encryption systems involve a coherent light field being propagated through some bulk optical systems, i.e., thin lenses and sections of free space. Of the encryption methods available, most examined involve the insertion of Random Phase Masks (RPMs) positioned in the input, such as Fourier, Fresnel, or some other Linear Canonical Transform domains. Significantly, most of these methods invariably result in a complex-valued output (i.e., encrypted image). In some cases, the RPMs in the optical encryption systems are implemented using Spatial Light Modulators to modulate the input data (amplitude, phase and polarization). Therefore, to capture the encrypted complex output field either Digital Holographic or Phase-Shift Interferometry is opted.

Digital images are often corrupted by several problems such as low resolution, blur, and noise, to name a few. Conventional image processing algorithms mostly work better in these cases and restore a good-quality image. Nevertheless, one of the major limitations of the conventional approaches is manual assistance i.e., human intervention, is required at all stages. Alternatively, deep learning has shown to be a promising approach as it identifies the problem and provides a tailor-made solution to it (adaptability) based on its prior learning.  

Hyperspectral Imaging  (HSI) is a new spectroscopy-based image analytical approach. This sensor gathers hundreds of bands of information in finer wavelengths, from visible to near-infrared regions. The photoreceptors in the human eye, on the other hand, respond only to the wavelength which corresponds to three primary colors i.e., Red, Green, and Blue. The collected HS data is presented in the form of a 3D cube which is represented as (x, y, λ) where x, y represent its spatial dimensions while λ gives its spectral information. HSI can also be used in in-vivo and ex-vivo cancer diagnosis as it provides the hidden details of the tissues that are not measurable with conventional techniques. It also provides an ionization-free diagnosis, thus safe for several investigations.