Ghost imaging (GI) illuminates an object with a sequence of light patterns and obtains the corresponding total echo intensities with a bucket detector. The correlation between the patterns and the bucket signals results in the image. Due to such a mechanism different from the traditional imaging methods, GI has received extensive attention during the past two decades. However, this mechanism also makes GI suffer from slow imaging speed and poor imaging quality. In previous work, each sample, including an illumination pattern and its detected bucket signal, was treated independently with each other. The correlation is therefore a linear superposition of the sequential data. Inspired by human’s speech, where sequential words are linked with each other by a certain semantic logic and an incomplete sentence could still convey a correct meaning, we here propose a different perspective that there is potentially a non-linear connection between the sequential samples in GI. We therefore built a system based on a recurrent neural network (RNN), called GI-RNN, which enables recovering high-quality images at low sampling rates. The test with MNIST’s handwriting numbers shows that, under a sampling rate of 1.28%, GI-RNN have a 12.58 dB higher than the traditional basic correlation algorithm and a 6.61 dB higher than compressed sensing algorithm in image quality. After trained with natural images, GI-RNN exhibits a strong generalization ability. Not only does GI-RNN work well with the standard images such as “cameraman”, but also it can recover the natural scenes in reality at the 3% sampling rate while the SSIMs are greater than 0.7.
Different from the traditional imaging methods using first-order interference, ghost imaging (GI) uses the second-order correlation, bringing many potential applications. On the other hand, GI has been suffering from low efficiency in image reconstruction due to a high sampling rate, which is a barrier for its application, especially when dealing with dynamic objects. The genetic algorithm (GA) can optimize the speckle sequence for an object and enable GI reconstruction with a few speckle patterns. However, the optimized speckle sequence of the GA usually loses the generalization and can only reconstruct the object being tested, making it far from suitable for handling a dynamic object. Here, we propose an improved method based on the GA, where we make two selection rules: the selective patterns more likely have a high response from the object, and meanwhile, the selected patterns tend to be linearly independent from each other. The optimized speckle sequence under these rules not only results in successful reconstruction but also preserves a generalization to a certain extent, enabling the GI to reconstruct the different states of the dynamic object at a low overall sampling rate. In the verification of the first frame, our method performs better based on the demonstration of various algorithms. In a demonstration of the dynamic object at 50% sampling rate, the reconstructed images are 2.1775 dB higher at 12 different frames on average in the peak signal-to-noise ratio than the random speckle sequence.
Ghost imaging is an unconventional imaging method, which has invoked many applications in various fields. However, it is still a major challenge to achieve high-fidelity high-resolution images at a sub-Nyquist sampling rate. Here, we present a ghost imaging method that illuminates an object with three directional Tetris-like patterns, which can greatly trade off the contradiction between the high resolution and high detection signal-to-noise ratio. As the projected patterns gradually shrink during the detection, the image is also gradually recovered from low to high resolution. In addition, this method can recover complex chromatic objects without any compromising image quality by adaptively abandoning unnecessary patterns at sampling rates well below the Nyquist limit. Meanwhile, the dynamic probing scheme has an excellent noise-removal capability. The simulation and experiment demonstrate that the sampling rate to recover a high-fidelity image is only [Formula: see text] for a scene of a [Formula: see text] duty cycle. For a very noisy scene whose peak signal–noise rate (PSNR) is 10.18 dB [the structural similarity index (SSIM) is 0.068], this scheme increases the PSNR to 18.63 dB [structural similarity index (SSIM) to 0.73]. Therefore, the proposed method may be useful for ghost imaging in the low sampling rate regime or complex chromatic objects reconstruction.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.