Liang Dong

Haze-induced image degradation poses a significant challenge to high-level vision tasks, particularly object detection in intelligent transportation systems. While recent dehazing methods have significantly improved conventional image quality metrics, they often fall short in enhancing downstream detection performance. Moreover, these methods lack comprehensive modeling of haze degradation in the frequency domain. To address these issues, this study investigates haze degradation from a wavelet-domain perspective and observes its scale-dependent characteristics. Specifically, haze-related information is predominantly concentrated in the low-frequency LL subbands, where different hierarchical levels capture degradation priors ranging from fine-grained light haze to coarse-grained dense haze. Building upon these insights, we propose D ² W-Net (Dynamic Wavelet Prior and Dual-Domain Learning Dehazing Network). The model employs Dynamic Wavelet Depthwise Separable Convolution (DyWConv) to enable content-adaptive feature encoding. It effectively unites the complementary strengths of wavelet and Fourier transforms under the MetaFormer architecture paradigm. Extensive experiments demonstrate that D ² W-Net achieves leading dehazing performance on synthetic datasets, attaining PSNR scores of 42.79 dB on SOTS-Indoor and 38.92 dB on SOTS-Outdoor, as well as superior results on real-world hazy images. More importantly, on the Foggy Cityscapes dense haze subset ( β =0.02), YOLOv8l equipped with our D ² W-Net achieves a 7.2 percentage point improvement in mAP@0.5 compared to the original hazy images. This result outperforms existing approaches and highlights its practical utility for robust perception in hazy traffic environments.

Infrared and visible image fusion improves the visual representation of scenes. Current deep learning-based fusion methods typically rely on either convolution operations for local feature extraction or Transformers for global feature extraction, often neglecting the contribution of multi-scale features to fusion performance. To address this limitation, we propose MRMAFusion, a nested connection model that relies on the multi-scale restoration-Transformer (Restormer) and multi-dimensional attention. We construct an encoder–decoder architecture on UNet++ network with multi-scale local and global feature extraction using convolution blocks and Restormer. Restormer can provide global dependency and more comprehensive attention to texture details of the target region along the vertical dimension, compared to extracting features by convolution operations. Along the horizontal dimension, we enhance MRMAFusion’s multi-scale feature extraction and reconstruction capability by incorporating multi-dimensional attention into the encoder’s convolutional blocks. We perform extensive experiments on the public datasets TNO, NIR and RoadScene and compare with other state-of-the-art methods for both objective and subjective evaluation.

Multispectral pedestrian detection, which combines visible and infrared images, has demonstrated significant advantages under various lighting and weather conditions, making it a highly focused research topic in recent years. The RGB-T modality essentially provides different descriptions of the same scene, encompassing both modality-specific and modality-consistent information. However, most existing approaches overlook the differences between these two types of information during feature fusion, leading to insufficient feature representation. To address this, we propose an illumination-aware octave fusion framework (OctNet) for RGB-T pedestrian detection. Specifically, we introduce an illumination-guided Octave Fusion module (IA-OctFuse), which utilizes frequency domain analysis to separate modality-complementary target features from redundant background features. Additionally, an illumination-aware mechanism is incorporated to adaptively balance the contributions of different modalities, producing highly discriminative RGB-T fused features. Then, a Multi-Head Dilated-convolution module (MHCE) is designed to deeply explore the spatial self-similarity of fused features, further enhancing feature representation. Extensive experiments and comparisons show that the proposed OctNet achieves state-of-the-art performance on publicly available KAIST and LLVIP pedestrian detection datasets.

This paper designs and implements an energy management system based on the Spring Boot framework. The system mainly includes three layers, which are the data collection layer, the business logic layer, and the display interface layer from bottom to top. The data collection layer is based on the RS-485 electrical standard and the MODBUS communication protocol. The two protocols connect all energy consumption monitoring points into a mixed topology communication network in the enterprise. The programs in the data collection layer poll each energy consumption monitoring point in the network to collect the data and transmit to the business logic layer. The business logic layer is developed on the basis of the Spring Boot framework and mainly includes two parts: the MySQL database and Tomcat server. In the MySQL database, the stored data are horizontally split according to the time column and stored in different data tables. The split of data reduces the load of a single data table and improves the query performance of the database. The Tomcat server is built into the Spring Boot framework to provide a basic environment for system operation. The Spring Boot framework is the core component of the system. It is responsible for collecting, storing, and analyzing data from energy consumption monitoring points, receiving and processing data requests from the display interface layer. It also provides standard data interfaces to the external programs. The display interface layer is developed on the basis of the Vue framework and integrated into the Spring Boot framework. The display layer combines an open-source visualization chart library called ECharts to provide users with a fully functional and friendly human–computer interaction interface. Through the calculation of hardware and software costs, considering the personnel cost in different regions, the total cost of the energy management system can be estimated. The cost of construction was approximately 210,000 USD in this paper. Since the system was actually deployed in a manufacturing company in December 2019, it has been operating stably for more than 600 days.