In the field of optics, it is essential to capture high-dimensional optical information in order to comprehend and characterize various targets in various scenes. This incorporates significant angles like irradiance, range, space, polarization, and stage. However, it can be quite challenging to locate a single system that is not only lightweight, portable, and cost-effective but also capable of efficiently gathering all of this data.
Enter compressive full-stack spectropolarimetric imaging (SPI), a strong procedure that incorporates a detached polarization modulator (PM) into a general imaging spectrometer. High-dimensional data can be obtained from insufficient measurements using this method. The main catch is the requirement for a remaking calculation to recuperate the 3D information blocks (x, y, and ) for each stirred-up boundary portraying the condition of electromagnetic radiation. Unfortunately, current reconstruction algorithms frequently have poor image quality and are sensitive to noise, and existing PMs are complicated and necessitate precise polarization calibration.
Fortunately, a group of Chinese researchers from Xi’an Jiaotong University have developed a novel solution. A passive spectropolarimetric modulation method with a single multiple-order retarder and polarizer has been developed, as detailed in Advanced Photonics Nexus. They propose a reconstruction algorithm that combines deep image prior and sparsity prior to create a unified forward imaging model for SPI. Best of all, their technique doesn’t require preparing information or exact polarization adjustment, and it can at the same time recreate the 3D information shapes while accomplishing self-alignment. The new method, known as DIP-SP, significantly raises the quality of the reconstructed images and is extremely resistant to noise.
The researchers created a single-shot SPI prototype by incorporating the simplest PM into a miniature snapshot imaging spectrometer to demonstrate the method’s viability and efficacy. They have demonstrated, through simulations and experiments, that their SPI scheme performs better than other methods. The fact that they are able to transform general spectral imaging systems into passive full-Stokes SPI systems without altering their inherent mechanisms is the beauty of their method.
The benefits of the Plunge SP strategy are complex. To begin, it makes it possible to encode and decode using the simplest passive PM scheme. Second, it doesn’t need any more training data and can be used for a variety of things. Third, it likewise disposes of the tedious quest for a scene-subordinate regularization boundary expected in different techniques. Fourth, the forward-imaging model makes the neural network’s output physically constrained and interpretable. Lastly, preliminary polarization calibration is acceptable, and accurate calibration of the measurement matrix is not required. By changing the measurement matrix, this method can also be used with other PM schemes.
This proposed SPI plan can be utilized in modern assessment and machine vision, where preview capacities are frequently required, and it additionally has possible applications in reconnaissance utilizing miniature automated elevated vehicles or ground vehicles. It can be easily adapted to a variety of commercial microscopies in the field of biomedical diagnosis, including fluorescence microscopy, confocal microscopy, ophthalmoscopes, endoscopes, and laryngoscopes.
Overall, the novel DIP-SP approach makes it possible to create exciting miniaturized SPI systems by integrating silicon photonic circuits or metasurfaces at the chip scale with free-space optical components. Its effortlessness, viability, and wide scope of uses make it a critical advance in the field of optical imaging.
More information: Feng Han et al, Deep image prior plus sparsity prior: toward single-shot full-Stokes spectropolarimetric imaging with a multiple-order retarder, Advanced Photonics Nexus (2023). DOI: 10.1117/1.APN.2.3.036009
