Abstract:Integrated optical power splitters are basic but indispensable on-chip devices in silicon photonics. They can be used either for power distribution or monitoring, or as the building blocks for more complex devices or circuits. Although different types of optical power splitters with different architectures have been proposed and demonstrated, devices that could work with arbitrary power splitting ratio in a large bandwidth without polarization dependence are still rare to be seen. In this paper, we propose and investigate an optical power splitter with adiabatically tapered waveguide structures on a thick silicon nitride platform, which could meet the requirement mentioned above. With optimized structural parameters obtained by three-dimensional finite-difference time-domain (3D-FDTD) simulation, the polarization dependence of different power splitting ratio gets almost eliminated for each specific working wavelength. In a broad wavelength range (1 340—1 800 nm), the insertion loss (IL) of the device is below 1 dB, and the variation of the power splitting ratio (PSR) can be controlled within ~±5% if compared with the targeted design value for 1 550 nm centered wavelength. Simple structure, relaxed critical dimensions, and good fabrication tolerance make this device compatible with the standard fabrication process in commercial silicon photonic foundries.

Abstract:Dater center interconnection has stimulated the development of short-reach optical communication transmission. To increase the capacity of the single sideband (SSB) system with direct-detection (DD), the twin-SSB system can double the system capacity without an extra optical modulator. Recently, the Stokes vector receiver (SVR) bridges the coherent detection and DD. A carrier can be sent along with the signal so that the SVR can utilize the reference carrier to conduct self-coherent detection and realize a linear complex optical detection, which is similar to the coherent receiver. In this paper, we propose a twin-SSB reception without optical SSB filters based on self-coherent detection. The proposed scheme greatly reduces the implementation complexity and has higher spectral efficiency (SE) compared with the traditional twin-SSB signal detection where two narrow-band optical filters and guard band are used for SSB signal extracting and inter crosstalk suppression. Meanwhile, the proposed twin-SSB signal detection with self-coherent detection also helps to relax the receiver digital signal processing (DSP), since the self-coherent detection reduces the inter crosstalk between the twin-SSB. The twin-SSB reception with self-coherent detection scheme makes it suitable for future high-speed short and medium-reach applications, such as the data center interconnect and metro area network.

Abstract:In order to mitigate the nonlinear effects of Mach-Zehnder modulator (MZM) on optical transmission signals in intensity modulation and direct detection (IM-DD) systems, a combined approach utilizing sinusoidal subcarrier modulation (SSM) and the Levenberg-Marquardt back propagation (LM-BP) neural network is proposed in this paper. The method employs a sine wave as the subcarrier to carry the 4 pulse amplitude modulation (PAM4) signals, aiming to equalize the distorted signals after MZM modulation. Subsequently, the LM-BP algorithm eliminates any remaining inter-symbol interference (ISI). This scheme uses sine wave modulation to solve the problem of additional ISI caused by triangular wave modulation. Furthermore, this combined approach simplifies the algorithm complexity compared to solely relying on a neural network equalizer. In this paper, the performance of SSM-LM-BP scheme is simulated and analyzed in IM-DD system. The results show that the joint scheme outperforms the triangular wave modulation scheme as well as the neural network algorithm after transmitting 50 Gbit/s PAM4 signals for 80 km without relays under the conditions of dispersion compensation, and the symbol error rate (SER) can be as low as 10-5.

Abstract:Models dedicated to building long-range dependencies often exhibit degraded performance when transferred to remote sensing images. Vision transformer (ViT) is a new paradigm in computer vision that uses multi-head self-attention (MSA) rather than convolution as the main computational module, with global modeling capabilities. However, its performance on small datasets is usually far inferior to that of convolutional neural networks (CNNs). In this work, we propose a small window SWinUNet (2SWUNet) for building extraction from high-resolution remote sensing images. Firstly, the 2SWUNet is trained based on swin transformer by designing a fully symmetric encoder-decoder U-shaped architecture. Secondly, to construct a reasonable U-shaped architecture for building extraction from high-resolution remote sensing images, different forms of patch expansion are explored to simulate up-sampling operations and recover feature map resolution. Then, the small window-based multi-head self-attention (W-MSA) is designed to reduce the computational and memory burden, which is more appropriate for the features of remote sensing images. Meanwhile, the pre-training mechanism is advanced to make up for the lack of decoder parameters. Finally, comparison experiments with other mainstream CNNs and ViTs validate the superiority of the proposed model.

Abstract:Metal active gas (MAG) welding is one of the widely applied welding techniques using argon and carbon dioxide as shielding gas. In response to the problem of welding halo and drag shadow during the image acquisition process of it, which makes it difficult to accurately extract the contour of the molten pool, this paper proposes a molten pool edge detection method that combines dark channel prior dehazing (DCPD) and improved single scale Retinex image enhancement algorithm. This method overcomes the problem of excessive edge noise in the original molten pool image and the difficulty in feature extraction caused by the dark part of the molten pool after DCPD processing. Through comparative experiments and ablation experiments, it has been shown that the algorithm proposed in this paper has significantly improved the enhancement effect and feature extraction effect, extracting accurate and complete molten pool contours.

Abstract:Blind watermarking plays a significant role in copyright protection and information hiding. To enhance the security of watermark image information, a dual encryption method is proposed, which combines block shuffling with Logistic chaos encryption using two sets of random numbers as keys. This dual encryption technique is applied to blind watermark images, utilizing a combination of discrete wavelet transform (DWT) and discrete cosine transform (DCT) as the embedding method. First, the watermark image is divided into small blocks, which are shuffled and rearranged using one set of random numbers as the key, while recording the shuffling sequence of block positions. Subsequently, the shuffled blocks are reassembled into an image, and another set of random numbers is used as the key for Logistic chaos encryption, rendering the watermark image chaotically invisible. Then, a double DWT is applied to the carrier image, followed by a DCT, allowing the encrypted watermark image to be embedded. This achieves the concealment of the blind watermark. Experimental results demonstrate that this method exhibits robustness, excellent invisibility, and secrecy.

Abstract:To address traffic congestion, this study improves MobileNetv2-you only look once version 4 (YOLOv4) target detection algorithm (MobileNetv2-YOLOv4-K++F) and introduces an embedded traffic information processing solution based on edge computing. We transition models initially designed for large-scale graphics processing units (GPUs) to edge computing devices, maximizing the strengths of both deep learning and edge computing technologies. This approach integrates embedded devices with the current traffic system, eliminating the need for extensive equipment updates. The solution enables real-time traffic flow monitoring and license plate recognition at the edge, synchronizing instantaneously with the cloud, allowing for intelligent adjustments of traffic signals and accident forewarnings, enhancing road utilization, and facilitating traffic flow optimization. Through on-site testing using the RK3399PRO development board and the MobileNetv2-YOLOv4-K++F object detection algorithm, the upgrade costs of this approach are less than one-tenth of conventional methods. Under favorable weather conditions, the traffic flow detection accuracy reaches as high as 98%, with license plate recognition exceeding 80%.

Abstract:The detection and localization of bone joint regions in medical X-ray images are essential for contemporary medical diagnostics. Traditional methods rely on subjective interpretation by physicians, leading to variability and potential errors. Automated bone joint detection techniques have become feasible with advancements in general-purpose object detection. However, applying these algorithms to X-ray images faces challenges due to the domain gap. To overcome these challenges, a novel framework called effective and efficient network (EAE-Net) is proposed. It incorporates a context augment module (CAM) to leverage global structural information and a ghost bottleneck module (GBM) to reduce redundant features. The EAE-Net model achieves exceptional detection performance, striking a balance between accuracy and speed. This advancement improves efficiency, enabling clinicians to focus on critical aspects of diagnosis and treatment.

Abstract:A reversible operation protocol is provided for a weak-measured quantum entanglement state. The evolution of weak values is studied under different parameter conditions. The weak values can be extracted from the entanglement state and the weak-measured quantum entanglement state can be revived to its initial state theoretically by weak measurement and reversibility operation respectively. We demonstrate the reversible operation protocol by taking Bell’s state as an example. The negativity is used to analyze the initial state, the weak-measured state and the reversed state in order to describe the evolution of quantum entanglement degree. Weak values are detected from the quantum entanglement state by weak measurement and the degree of the weak-measured quantum entanglement state can be revived to its initial state through reversible operation. The information of quantum entanglement state would be extracted from weak values detected in the process of the scheme.