As the Neural Radiance Fields (NeRF) have achieved significant success in view synthesis, some works attempt to apply volume rendering to 3D reconstruction task. However, these methods use positional encoding with uniform frequency for the whole scenario during training, ignoring different regions of the scene and different training stage. Therefore, we aim to design a frequency-adaptive positional encoding. The adaptivity here refers to two points: first, it is adaptive to scene regions, meaning different regions of each scene have different frequency of positional encoding. Second, it is adaptive to the training process, meaning frequency changes during training to adapt to the convergence of the network. To achieve this adaptivity, we use a neural network to learn the frequency field of scenes, that is, using the network to predict the required frequency for each point during the training process. The introduction of the frequency field allows each scene to obtain an adaptive positional encoding, enabling the neural network to learn the geometric information of scenes in a more flexible and efficient manner. Experimental results show that our approach produces more accurate surfaces compared to baseline methods, especially in scenes with complex geometry.

Enhancing physical layer security through the deployment of intelligent reflecting surfaces (IRS) is an emerging and significant research direction in 6G wireless networks. Existing methods mainly focus on optimizing the transmit beamforming of the access point (AP) and the phase shift of the IRS to maximize the secrecy rate, while neglecting the optimization of the IRS amplitude. We propose a joint op-timization framework that simultaneously optimizes the IRS amplitude, IRS phase shifts, and AP beamforming to enhance the secrecy rate. Due to the problem’s non-convexity, exist-ing alternating optimization methods face high challenges, particularly for large-scale IRS systems and high-power sce-narios. To tackle these issues, we propose a competition mechanism-based multi-strategy swarm optimization algo-rithm, which aims to better balance global exploration and local exploitation capabilities, thereby achieving more opti-mal IRS configurations. Simulations demonstrate that jointly optimizing IRS amplitude, phase, and AP beamforming sig-nificantly improves secrecy rate over phase-only beamforming methods.

When estimating the direction of arrival (DOA) of wideband signals from multiple sources, the performance of sparse Bayesian methods is influenced by the frequency bands occupied by signals in different directions. This is particularly true when multiple signal frequency bands overlap. Message passing algorithms (MPA) with Dirichlet process (DP) prior can be employed in a sparse Bayesian learning (SBL) framework with high precision. However, existing methods suffer from either high complexity or low precision. To address this, we propose a low-complexity DOA estimation algorithm based on a factor graph. This approach introduces two strong constraints via a stretching transformation of the factor graph. The first constraint separates the observation from the DP prior, enabling the application of the unitary approximate message passing (UAMP) algorithm for simplified inference and mitigation of divergence issues. The second constraint compensates for the deviation in estimation angle caused by the grid mismatch problem. Compared to state-of-the-art algorithms, our proposed method offers higher estimation accuracy and lower complexity.

Retinal images play an essential role in the early diagnosis of ophthalmic diseases. Automatic segmentation of retinal vessels in color fundus images is challenging due to the morphological differences between the retinal vessels and the low-contrast background. At the same time, automated models struggle to capture representative and discriminative retinal vascular features. To fully utilize the structural information of the retinal blood vessels, we propose a novel deep learning network called Pre-Activated Convolution Residual and Triple Attention Mechanism Network (PCRTAM-Net). PCRTAM-Net uses the pre-activated dropout convolution residual method to improve the feature learning ability of the network. In addition, the residual atrous convolution spatial pyramid is integrated into both ends of the network encoder to extract multiscale information and improve blood vessel information flow. A triple attention mechanism is proposed to extract the structural information between vessel contexts and to learn long-range feature dependencies. We evaluate the proposed PCRTAM-Net on four publicly available datasets, DRIVE, CHASE_DB1, STARE, and HRF. Our model achieves state-of-the-art performance of 97.10%, 97.70%, 97.68%, and 97.14% for ACC and 83.05%, 82.26%, 84.64%, and 81.16% for F1, respectively.