Low-Light Image Enhancement (LLIE) is crucial for improving image quality and visual analysis. This study proposes MLSwinTNet, an LLIE network based on the Swin Transformer. The core MLSwinT module adopts a UNet-like design, integrating a dual-branch feature extraction module and a multi-level feature interaction module. The former combines CNN and Swin Transformer to efficiently capture both local and global features, while the latter optimizes detail and color restoration by adjusting and fusing feature levels. Experimental results demonstrate that MLSwinTNet outperforms state-of-the-art methods on standard datasets, achieving a PSNR gain of 27.33 dB and an SSIM score of over 0.96, providing a more competitive solution for LLIE.

Image dehazing aims to restore content in images degraded by atmospheric haze. However, this task remains highly challenging due to the non-uniform distribution of haze across scenes and the inherent complexity of image content itself. Current mainstream approaches predominantly employ Convolutional Neural Networks (CNNs) for end-to-end blind dehazing training. However, these methods exhibit strong dependence on training data and struggle to effectively model the inherent physical relationship between haze concentration and scene depth. To address this limitation, we propose a Depth-guided Sparse Attention Network (DGSAN) for single-image dehazing. Our method first accurately estimates the scene depth map, embedding these depth features within a unified dehazing framework. Subsequently, through a depth-guided sparse attention mechanism, it dynamically fuses depth features with image features, thereby guiding the network to focus on suboptimal dehazed areas to improve the network performance. Experimental results demonstrate that the proposed method exhibits superior performance on both synthetic and real-world image datasets.

Enhancement of low-light images is a low-level visual task aimed at improving the quality of images captured under low-light conditions. In this study, a low-light image enhancement algorithm is proposed by compensating for the reflection loss of illumination components obtained through object reflection. Specifically, the algorithm first utilizes Gaussian filtering to process the value component (V) of the image, separating the illumination component from the reflection component. Then, through a illumination compensation strategy, the illumination component is processed and combined with the reflectance component to synthesize the enhanced value component (V). Finally, an adaptive global balance strategy is applied to optimize the enhanced value component (V) to ensure that the resulting image appears more natural and conforms to human visual perception habits. Experimental results demonstrate the effectiveness and superiority of our method compared to existing traditional processing algorithms and deep learning methods, showing excellent performance in enhancing dark details of images and maintaining natural colors.

Image dehazing is a fundamental low-level vision task aimed at recovering clear visual scenes from images affected by haze. Addressing the haze effects in real outdoor environments, we propose a simple yet effective image prior: haze color prior. Leveraging this prior knowledge, we devise a computational model for estimating the coarse-level transmission map and obtain a refined transmission map through an adaptive gamma correction method based on the dark channel prior. Subsequently, employing our proposed two-stage filtering strategy, appropriate local atmospheric light can be acquired. Finally, integrating with the atmospheric scattering model, we can restore high-quality haze-free images. Experimental results prove that compared to other existing advanced methods, this algorithm excels in dehazing, enhancing image contrast, and preserving details.

Low-light image enhancement is a significant and challenging task in the field of computer vision. This paper presents a method combining enhanced anisotropic diffusion techniques with Retinex theory for low-light image enhancement. Initially, the algorithm employs the HSV color space to decompose the image, with a focus on processing the luminance component. In the crucial illumination estimation step, an improved anisotropic diffusion technique is proposed. This technique adaptively modulates the diffusion process based on the image’s gradient information, effectively capturing both local and global illumination details and ensuring precise and smooth illumination estimation. Subsequently, the paper proposes a global illumination balancing strategy to maintain the natural light and dark relationships within the image. Finally, an adaptive illumination enhancement function is developed, significantly improving the visibility in dark areas while preserving the essential details of the image. Experimental validation demonstrates that this method outperforms existing approaches in low-light image enhancement, paving new pathways for applications in low-light vision.

Low-light image enhancement is a crucial low-level visual task, essential for improving image quality, enhancing visual effects, and facilitating higher-level image analysis and processing. This paper introduces an improved multi-scale Retinex algorithm for enhancing low-light images. Initially, the low-light image is converted to the HSV color space, and the improved multi-scale Retinex is applied to the V (brightness) channel to obtain the illumination component and the optimized reflection component, the latter serving as the enhanced brightness channel. Subsequently, a brightness compensation strategy based on the estimated illumination component is proposed for adaptive adjustment of the enhanced brightness channel. Finally, by restoring the brightness values in high-luminance areas, the enhanced brightness channel V′ is formed, combined with the H and S channels, and converted back to the RGB color space to produce the final enhanced image. Experimental results show that this method is effective for low-light images with uniform or non-uniform lighting, significantly improving image quality and providing an excellent visual experience. The code is hosted in the GitHub repository at https://github.com/xinxin6809/IMR-ABC.

The movement of nodes in opportunistic networks exhibits characteristics of clustering and regularity. Consequently, routing algorithms based on communities have become a current research hotspot. However, existing community-based routing algorithms do not comprehensively analyze both the overlap of node communities and the impact of neighboring nodes on message transmission. To address this issue, this paper proposes a novel opportunistic routing algorithm called CWON, based on overlapping communities and communication willingness. First, we utilize the PercoMCV method to partition overlapping communities. Subsequently, we define the concept of communication willingness and design the CWON algorithm based on this concept. The CWON algorithm effectively addresses the problem of overlapping community partitioning and measures the importance of different neighboring nodes in message transmission. Our simulation results demonstrate that the CWON algorithm significantly enhances the success rate of message delivery while reducing routing overhead.

Opportunistic network nodes exhibit social attributes, and existing community routing algorithms are currently designed for situations where the community structure remains fixed and do not comprehensively analyze the impact of node location information on data forwarding. Over time, the community division results do not match the current network topology structure, and it becomes challenging to select appropriate relay nodes for forwarding. In order to solve this problem, this paper proposes a community routing based on node location information-CRLI. Firstly, the communities are partitioned based on node interaction information, and then the regional affiliation and regional connectivity are defined. Based on these, the CRLI algorithm is designed to comprehensively analyze the influence of node location, movement direction, and dynamic changes in community structure on data forwarding. The experimental results show that the CRLI algorithm can effectively improve the message delivery rate and reduce overhead.

This paper describes a digital method of reducing timing mismatch effects in time-interleaved ADCs used in ATE systems: we use cross-correlation among channel ADC outputs to detect channel timing skew, and make successive-approximation adjustments to our proposed linear-phase-digital delay filter to compensate for the timing skew. Simulation results validate the effectiveness of the proposed method. We found that using multitone input signals with cross-correlation of outputs provided a more robust way of detecting timing skew than using a singletone input signal. Since our proposed approach is fully digital, it is reliable, and suitable for fine CMOS implementation.