Single-pixel imaging based on self-supervised conditional mask classifier-free guidance.

Reconstructing high-quality images at a low measurement rate is a pivotal objective of Single-Pixel Imaging (SPI). Currently, deep learning methods achieve this by optimizing the loss between the target image and the original image, thereby constraining the potential of low measurement values. We employ conditional probability to ameliorate this, introducing the classifier-free guidance model (CFG) for enhanced reconstruction. We propose a self-supervised conditional masked classifier-free guidance (SCM-CFG) for single-pixel reconstruction. At a 10% measurement rate, SCM-CFG efficiently completed the training task, achieving an average peak signal-to-noise ratio (PSNR) of 26.17 dB on the MNIST dataset. This surpasses other methods of photon imaging and computational ghost imaging. It demonstrates remarkable generalization performance. Moreover, thanks to the outstanding design of the conditional mask in this paper, it can significantly enhance the accuracy of reconstructed images through overlay. SCM-CFG achieved a notable improvement of an average of 7.3 dB in overlay processing, in contrast to only a 1 dB improvement in computational ghost imaging. Subsequent physical experiments validated the effectiveness of SCM-CFG.

Single photon counting compressive imaging using a generative model optimized via sampling and transfer learning.

Single photon counting compressive imaging, a combination of single-pixel-imaging and single-photon-counting technology, is provided with low cost and ultra-high sensitivity. However, it requires a long imaging time when applying traditional compressed sensing (CS) reconstruction algorithms. A deep-learning-based compressed reconstruction network refrains iterative computation while achieving efficient reconstruction. This paper proposes a compressed reconstruction network (OGTM) based on a generative model, adding sampling sub-network to achieve joint-optimization of sampling and generation for better reconstruction. To avoid the slow convergence caused by alternating training, initial weights of the sampling and generation sub-network are transferred from an autoencoder. The results indicate that the convergence speed and imaging quality are significantly improved. The OGTM validated on a single-photon compressive imaging system performs imaging experiments on specific and generalized targets. For specific targets, the results demonstrate that OGTM can quickly generate images from few measurements, and its reconstruction is better than the existing compressed sensing recovery algorithms, compensating defects of the generative models in compressed sensing.

Deep-learning-based single-photon-counting compressive imaging via jointly trained subpixel convolution sampling.

The combination of single-pixel-imaging and single-photon-counting technology can achieve ultrahigh-sensitivity photon-counting imaging. However, its applications in high-resolution and real-time scenarios are limited by the long sampling and reconstruction time. Deep-learning-based compressive sensing provides an effective solution due to its ability to achieve fast and high-quality reconstruction. This paper proposes a sampling and reconstruction integrated neural network for single-photon-counting compressive imaging. To effectively remove the blocking artefact, a subpixel convolutional layer is jointly trained with a deep reconstruction network to imitate compressed sampling. By modifying the forward and backward propagation of the network, the first layer is trained into a binary matrix, which can be applied to the imaging system. An improved deep-reconstruction network based on the traditional Inception network is proposed, and the experimental results show that its reconstruction quality is better than existing deep-learning-based compressive sensing reconstruction algorithms.

Full Stokes single photon compressive polarization imaging.

We demonstrate a full Stokes polarization imaging system based on compressed sampling and single photon counting. The control and synchronization counting module based on field-programmable gate array is specially developed to control the rotation stage for polarization imaging at different directions. Additionally, it can load the binary random matrix into a digital micro-mirror device controller for each measurement and count the single photon pulse output from the photon counting photomultiplier tube simultaneously. The system can realize high-sensitivity single photon compressive imaging of the target under different polarization directions. On this basis, the high-quality Stokes parameter images and the angle of the linear polarization image can be obtained. The experimental results show that the polarization information can be reconstructed at a very low sampling ratio.

A binary sampling Res2net reconstruction network for single-pixel imaging.

The traditional algorithm for compressive reconstruction has high computational complexity. In order to reduce the reconstruction time of compressive sensing, deep learning networks have proven to be an effective solution. In this paper, we have developed a single-pixel imaging system based on deep learning and designed the binary sampling Res2Net reconstruction network (Bsr2-Net) model suitable for binary matrix sampling. In the experiments, we compared the structural similarity, peak signal-to-noise ratio, and reconstruction time using different reconstruction methods. Experimental results show that the Bsr2-Net is superior to several deep learning networks recently reported and closes to the most advanced reconstruction algorithms.

Photon-Counting Underwater Optical Wireless Communication for Reliable Video Transmission Using Joint Source-Channel Coding Based on Distributed Compressive Sensing.

To achieve long-distance underwater optical wireless communication, a single photon detector with single photon limit sensitivity is used to detect the optical signal at the receiver. The communication signal is extracted from the discrete single photon pulses output from the detector. Due to fluctuation of photon flux and quantum efficiency of photon detection, long-distance underwater optical wireless communication has the characteristics that the link is easily interrupted, the bit error rate is high, and the burst error is large. To achieve reliable video transmission, a joint source-channel coding scheme based on residual distributed compressive video sensing is proposed for the underwater photon counting communication system. Signal extraction from single photon pulses, data frame and data verification are specifically designed. This scheme greatly reduces the amount of data at the transmitter, transfers the computational complexity to the decoder in receiver, and enhances anti-channel error ability. The experimental results show that, when the baud rate was 100 kbps and the average number of photon pulses per bit was 20, the bit error rate (BER) was 0.0421 and video frame could still be restored clearly.

Measurement Matrix Construction for Large-area Single Photon Compressive Imaging.

We have developed a single photon compressive imaging system based on single photon counting technology and compressed sensing theory, using a photomultiplier tube (PMT) photon counting head as the bucket detector. This system can realize ultra-weak light imaging with the imaging area up to the entire digital micromirror device (DMD) working region. The measurement matrix in this system is required to be binary due to the two working states of the micromirror corresponding to two controlled elements. And it has a great impact on the performance of the imaging system, because it involves modulation of the optical signal and image reconstruction. Three kinds of binary matrix including sparse binary random matrix, m sequence matrix and true random number matrix are constructed. The properties of these matrices are analyzed theoretically with the uncertainty principle. The parameters of measurement matrix including sparsity ratio, compressive sampling ratio and reconstruction time are verified in the experimental system. The experimental results show that, the increase of sparsity ratio and compressive sampling ratio can improve the reconstruction quality. However, when the increase is up to a certain value, the reconstruction quality tends to be saturated. Compared to the other two types of measurement matrices, the m sequence matrix has better performance in image reconstruction.

Adaptive Single Photon Compressed Imaging Based on Constructing a Smart Threshold Matrix.

We demonstrate a single-photon compressed imaging system based on single photon counting technology and compressed sensing theory. In order to cut down the measurement times and shorten the imaging time, a fast and efficient adaptive sampling method, suited for single-photon compressed imaging, is proposed. First, the pre-measured rough images are transformed into sparse bases as a priori information. Then a smart threshold matrix is designed by using large sparse coefficients of the rough image in sparse bases. The adaptive measurement matrix is obtained by modifying the original Gaussian random matrix with the specially designed threshold matrix. Building the adaptive measurement matrix requires only one level of sparse representation, which means that adaptive imaging can be achieved quickly with very little computation. The experimental results show that the reconstruction effect of the image measured using the adaptive measurement matrix is obviously superior than that of the Gaussian random matrix under different measurement times and different reconstruction algorithms.

Large-area single photon compressive imaging based on multiple micro-mirrors combination imaging method.

We demonstrate a single photon compressive imaging system with the image plane up to the entire digital micro-mirror device (DMD) work area. A parallel light source is designed to reduce the influence of light scattering on imaging resolution and a photon counting photomultiplier tube (PMT) with a large photosensitive area is used to effectively collect light reflected from the full screen of DMD. A control and counting circuit, based on Field-Programmable Gate Array (FPGA), is developed to load binary random matrix into the DMD controller for each measurement, and to count single-photon pulse output from PMT simultaneously. To reduce imaging time and huge memory occupation for image reconstruction, a multiple micro-mirrors combination imaging method is proposed. The signal-to-noise ratio and detection limit of the imaging system is theoretically deduced. Theoretical analysis and experimental results show that micro-mirrors combination imaging method is more suitable for faster imaging in a weaker-light-level environment. In order to achieve high imaging quality, the size of the combined pixels and the average time of each measurement should be moderate, so that the impact of Poisson shot noise is minimized.

Effects of donor and acceptor's fluorescence lifetimes on the method of applying Förster resonance energy transfer in STED microscopy.

Förster resonance energy transfer (FRET) probes being used to improve the resolution of stimulated emission depletion (STED) microscopy are numerically discussed. Besides the FRET efficiency and the excitation intensity, the fluorescence lifetimes of donor and acceptor are found to be another key parameter for the resolution enhancement. Using samples of FRET pairs with shorter donor lifetime and longer acceptor lifetime enhances the nonlinearity of the donor fluorescence, which leads to an increased resolution. The numerical simulation shows that a double resolution improvement of STED microscopy can be achieved by using Cy3-Atto647N samples when compared with that of using standard Cy3-only samples.

High-speed quantum-random number generation by continuous measurement of arrival time of photons.

We demonstrate a novel high speed and multi-bit optical quantum random number generator by continuously measuring arrival time of photons with a common starting point. To obtain the unbiased and post-processing free random bits, the measured photon arrival time is converted into the sum of integral multiple of a fixed period and a phase time. Theoretical and experimental results show that the phase time is an independent and uniform random variable. A random bit extraction method by encoding the phase time is proposed. An experimental setup has been built and the unbiased random bit generation rate could reach 128 Mb/s, with random bit generation efficiency of 8 bits per detected photon. The random numbers passed all tests in the statistical test suite.

Multi-bit quantum random number generation by measuring positions of arrival photons.

We report upon the realization of a novel multi-bit optical quantum random number generator by continuously measuring the arrival positions of photon emitted from a LED using MCP-based WSA photon counting imaging detector. A spatial encoding method is proposed to extract multi-bits random number from the position coordinates of each detected photon. The randomness of bits sequence relies on the intrinsic randomness of the quantum physical processes of photonic emission and subsequent photoelectric conversion. A prototype has been built and the random bit generation rate could reach 8 Mbit/s, with random bit generation efficiency of 16 bits per detected photon. FPGA implementation of Huffman coding is proposed to reduce the bias of raw extracted random bits. The random numbers passed all tests for physical random number generator.

The resolution estimation of wedge and strip anodes.

A new resolution estimation method for wedge and strip anode based on the single photon imaging configuration is provided. The limiting resolution estimation equation is deduced theoretically according to the threshold principle. The relation between the charge cloud and the covered electrodes is discussed, and the equivalent diameter number is calculated. The resolution equation for the position deviation amplitude or FWHM is provided if noise exists. The relation between the position deviation amplitude and the total charge deviation amplitudes is discussed. The constancy of the position deviation amplitudes versus positions is provided. The results calculated from these equations are discussed. According to the equations, it is indicated that the spatial resolution is affected by the detection system configuration and noise. These conclusions may be useful for the designing and performance improvement of future photon imagers.

Continuous measurement of the arrival times of x-ray photon sequence.

In order to record x-ray pulse profile for x-ray pulsar-based navigation and timing, this paper presents a continuous, high-precision method for measuring arrival times of photon sequence with a common starting point. In this method, a high stability atomic clock is counted to measure the coarse time of arrival photon. A high resolution time-to-digital converter is used to measure the fine time of arrival photon. The coarse times and the fine times are recorded continuously and then transferred to computer memory by way of memory switch. The pulse profile is obtained by a special data processing method. A special circuit was developed and a low-level x-ray pulse profile measurement experiment system was setup. The arrival times of x-ray photon sequence can be consecutively recorded with a time resolution of 500 ps and the profile of x-ray pulse was constructed. The data also can be used for analysis by many other methods, such as statistical distribution of photon events per time interval, statistical distribution of time interval between two photon events, photon counting histogram, autocorrelation and higher order autocorrelation.

The estimation of charge footprint size of ultraviolet photon counting imaging detector with induction readout.

The structure and work principle of an ultraviolet photon counting imaging detector based on wedge and strip anode with induction readout mode are introduced. Two methods of estimating the charge footprint size are presented. One way is theoretical calculation and simulation. The physical course of electrons is simulated from the microchannel plate output side to the readout anode. The calculated results show that the final charge footprint size is sensitive to the thickness of ceramic and not sensitive to the charge footprint size on the Ge layer. The other way is experimental image estimation. The final charge footprint size can be estimated according to the position where the light line of resolution board image begins to bend. Both methods show that the charge footprint size is sensitive to the ceramic substrate. The two methods are simple and effective for estimation of charge footprint size of photon counting imaging detector with induction readout.