Existing CNN-based and Transformer-based methods have demonstrated remarkable performance in low-level visual tasks, including image deblurring. These methods generally capture spatial features only in a single way, such as by stacking blocks of CNNs and Transformers, resulting in inadequate utilization of spatial context. To address this issue, we propose a new feature aggregation scheme for image deblurring, named Omni-Deblurring. The core of our omni-deblurring is the omni-range context block, which enables explicitly aggregating the local-range, regional-range, and global-range features in a compact manner. With this design, it can bring a wider receptive field for modeling the contextual features. Extensive experiments on synthetic and real-world blurry datasets demonstrate the effectiveness of our proposed method in both quantitative and qualitative evaluations. Furthermore, the quality of our deblurring model is evaluated in the task of object detection, and the mean Average Precision (mAP) metric increases 10% across all classes compared with other deblurring models. Code is available at https://github.com/yaowli468/Omni-Deblurring.

Three-dimensional (3D) recovery through scattering media remains a central challenge in optical imaging. The medium randomly scatters the optical field, mixing and obscuring information and rendering conventional decoding strategies ineffective. To address this, we propose polarization to impose physical constraints on intensity recovery, as build a key bridge between descattering imaging and 3D recovery. A polarization-guided network architecture is developed that directly incorporates the polarization characteristics of the scattered field into the network as differentiable physical constraints. By combining learned feature extraction with explicit physical computation, this enhances the accuracy of physical information recovery, thereby laying a reliable foundation for subsequent 3D recovery. Experimental results show that the proposed method yields an improvement in the joint recovery of intensity and polarization information and successfully reconstructs 3D recovery under random scattering conditions.

This study introduces a novel imaging technology designed to address the critical challenges of high integration, wide field of view (FOV), and high-resolution detection in optoelectronic systems. The proposed approach leverages ultra-small pixels and a unique “single-eye + compound-eye” architecture, combining a concentric spherical lens for wide-FOV light collection with a distributed camera array for segmented imaging and stitching of the Petzval image plane. This design enables high-resolution imaging across a large area while maintaining compact system dimensions. The ultra-small-pixel large-format detector excels in capturing fine details, and by applying linear system theory, the technology achieves significant reductions in system size and complexity without compromising performance. Experimental validation shows that the imaging system achieves a modulation transfer function near the diffraction limit at 500 lp/mm, with a root-mean-square spot size consistently below 1 $\mathrm{\mu }$m. Additionally, the system delivers an angular resolution of 25 $\mathrm{\mu }$rad and distortion-free imaging over a ${61.5}^{\circ }\times {55}^{\circ }$ FOV.

Strong light exceeding the saturation intensity threshold and the dynamic range of imaging systems leads to overexposure and severe image degradation. Current approaches generally rely on intricate hardware setups, sophisticated algorithms, or multi-exposure strategies, which often bring about bulky systems, high computational costs, and limited adaptability. In this work, we propose what is believed to be a novel method to simultaneously enhance the saturation intensity and the dynamic range of a polarization imaging system by introducing a low-cost polarization modulation aperture (PMA). The core principle is to reduce the intensity reaching the camera by modulating the proportion of polarization components in the incoming light. This allows the actual intensity to be retrieved from a single capture via simple linear demodulation. By directly integrating the thin PMA directly into the lens assembly, our method avoids elaborate hardware alterations or computationally intensive post-processing. Moreover, the attenuation level can be adjusted by configuring the area ratios and polarization modulation directions of the sub-apertures within the PMA, ensuring optimized performance across diverse scenarios. Experiments show that, with a customized PMA using prior scene information, the prototype improves the saturation intensity by 680% and the effective dynamic range, meaning the system’s recoverable intensity range, by 17 dB in indoor strong light. In contrast, a generic PMA without prior information yields 200% and 6 dB gains in non-prior scenarios. Our method thus provides a compact, efficient, and cost-effective solution for achieving single-shot, high-quality polarization imaging under strong light conditions, particularly in environments with distinct polarization characteristics.