In recent years, with the development of micro-nano optics research, lensless imaging systems based on micro-nano structured diffusers have attracted extensive attention. Unlike the point-to-point imaging method used in traditional imaging systems, the lensless imaging system replaces the traditional lens with diffusers, directly recording the object's encoded pattern on the sensor. The object's information is then recovered through postprocessing algorithms. Due to the lack of a lens, lensless imaging systems have lower manufacturing costs, and can greatly reduce the size and weight of the system, making them suitable for small and portable devices. Recently, using disordered scattering media as imaging lenses or encoding diffusers, significant progress has been made in scattering media imaging, super-resolution imaging and wide-field holography.
However, in order to use disordered scattering media as imaging lenses, time-consuming and cumbersome characterization work is inevitable, especially for time-varying scattering media. In addition, due to the limited modulation capability of disordered scattering media on the optical field, traditional scattering imaging systems are insensitive to wavelength and polarization, huge challenges exist in achieving multidimensional imaging such as polarization, wavelength, depth, and two-dimensional space simultaneously.
Metasurface is a planar structure composed of tiny structural units, sizes of uints are typically in micro/nano scale. It can be designed to modulate light field in a specific way to control propagation and scattering characteristics according to the application requirements. Owing to its strong optical field modulation capability, metasurface can be used as a random coding diffuser to encode the polarization, wavelength, depth and two-dimensional spatial position distribution of the incident light, enabling multidimensional computational imaging. Although significant progresses have been made in the design and manufacturing of metasurfaces, achieving efficient mass production and replication remains a critical scientific and technological challenge in this field.
To address the problems, group from Research Center of Vector Optical Field, Institute of Optics and Electronics, Chinese Academy of Sciences reported a liquid crystal metasurface diffuser based on geometric phase. Owing to the feature of prior wavefront modulation, spatial scattering characteristics of metasurfaces are more stable and controllable. As an pre-designed encoding diffuser, time-consuming pre-characterization can be skipped. In addition, according to the principle of geometric phase and scattering medium memory effect, liquid crystal diffuser exhibits sensitivities in polarization, wavelength, and depth, while being insensitive to two-dimensional planer space. Therefore, it can encode five-dimensional (5D) information of the imaging target in form of point spread function (PSF). Finally, compared to other metasurfaces made of dielectric or metal, the manufacture of the liquid crystal metasurface is simpler as well as more conducive to large-scale processing and production. The relevant results were published in Photonics Research, Volume 3, No. 11, 2023 (Yunsong Lei, Qi Zhang, Yinghui Guo, Mingbo Pu, Fang Zou, Xiong Li, Xiaoliang Ma, Xiangang Luo. Snapshot multi-dimensional computational imaging through a liquid crystal diffuser[J]. Photonics Research, 2023, 11(3): B111).
As shown in Figure 1, this lensless system encodes the polarization, wavelength, depth and spatial information of objects by a liquid crystal metasurface diffuser. By using a point light source which shares the same polarization, wavelength, and depth as the imaged object and passes through the same optical system, the corresponding PSF can be obtained. Then, the deconvolutional algorithm serves as a decoding method to restore the 5D information encoded in the speckle patterns.
Fig.1 (a) Schematic diagram of snapshot imaging system, (b) Imaging objects with different five-dimensional information have unique point spread functions, (c) Deconvolve the speckle pattern with different point spread functions to obtain the corresponding information, (d) Combine the different information to obtain the complete image.
The experimental process and the results are depicted in Figure 2, where two objects with different wavelengths were projected onto distinct areas by a digital mirror device (DMD) projector. Then, two objects were separated by mirrors, and the polarization state of each object was altered by using a combination of polarizers and quarter-wave plates with different rotation angles. Subsequently, the separated objects were combined using a beam combiner and transmitted through the liquid crystal diffuser to generate a mixed speckle pattern. Next, a pixel at the corresponding position of the projector was lightened asynchronously, which have the same wavelengths as the previous objects. As passing through the same path, the obtained speckles are the corresponding point spread functions. Finally, the relevant object information was obtained by deconvolving the speckle pattern with corresponding PSFs.
Fig. 2 (a) Schematic diagram of the five-dimensional imaging system, (b) The superimposed speckle pattern image, (c) and (d) Deconvolution of different point spread functions and speckle patterns can obtain corresponding information.
The results suggest a cost-effective and compact approach to 5D imaging, which has potential applications in material identification and classification, as well as in biomedicine and various industrial fields.
