Ligen Shi

High-quality light field (LF) acquisition involves a trade-off between spatial and angular resolution. Hybrid camera systems offer a practical solution but present challenges for dense reconstruction due to sparse angular sampling and complex occlusions. To address these issues, this paper proposes a geometry-aware implicit neural representation (INR) framework for LF reconstruction. Distinct from traditional discrete representations or generic ray-space embeddings, we introduce a compact, continuous representation that combines a radiance field with a geometry module, where scalar disparity provides an ideal geometric interpretation and a ray-displacement field serves as the practical realization for occlusion-aware reconstruction. This framework leverages the global epipolar geometry of the 2D LF grid while utilizing the displacement field to correct local geometric inconsistencies caused by occlusions and non-Lambertian effects. By coupling these fields with a radiance network, our method enables end-to-end differentiable optimization from sparse, multi-resolution inputs without relying on large-scale external training datasets. Experiments on hybrid camera data fusion and spatial-angular super-resolution tasks demonstrate that our approach preserves high-frequency details and geometric consistency.

Accurate structural analysis of coated fuel particles is essential for assessing the performance and quality of advanced nuclear fuels. X-ray computed tomography (CT) provides a non-destructive means to visualize the multilayer structure of TRISO fuel particles. However, accurate layer segmentation remains challenging due to low interlayer contrast, imaging noise, and limited labeled data. This paper proposes a physics-guided multistage segmentation network (PMSNet) for automatic multilayer segmentation of TRISO fuel particle CT images. PMSNet integrates prior knowledge of the particle’s spherical geometry and coating hierarchy into a deep learning framework, and consists of three key components: axial-aware position encoding with layer-specific spatial priors, a multiscale perturbation strategy to enhance generalization, and a prior-guided conditional fusion mechanism to enforce interlayer consistency. Experiments on synthetic and real TRISO CT datasets demonstrate that PMSNet achieves mean Dice coefficients above 0.98 and mean IoU values exceeding 0.97 across multiple coating layers. Compared with state-of-the-art methods, PMSNet improves Dice scores by 3–8 percentage points on challenging layers and achieves boundary F1 scores close to 1.0 for key interfaces. While the primary focus of this study is TRISO fuel particle segmentation, additional evaluations are performed to examine the general applicability of the proposed framework. Results on medical CT datasets further confirm its robustness and adaptability to radiation-based imaging tasks. The proposed method contributes a physics-based AI segmentation framework, enables accurate and fully automatic multilayer segmentation of TRISO fuel particles, and provides a practical solution for more efficient and reliable non-destructive nuclear fuel inspection.

High-fidelity rendering of complex scenes remains challenging for 3D Gaussian Splatting (3DGS), particularly under large viewpoint variations and in regions dominated by high-frequency structures, where aliasing and geometric inconsistencies frequently arise. To address these issues, we propose a wavelet-guided and depth-aware optimization framework that enhances frequency preservation and structural consistency in 3DGS-based computational imaging. A multi-scale wavelet decomposition is integrated to extract discriminative frequency components, which explicitly guide the reconstruction of high-frequency details during splat optimization. In parallel, a depth-aware weighting mechanism is incorporated into the rendering pipeline to adaptively modulate the contributions of spatially adjacent Gaussian primitives, improving geometric stability and mitigating artifacts caused by occlusion or rapid depth transitions. Furthermore, we introduce a high-frequency–weighted hybrid loss function based on L ₁ distance and SSIM, enabling the network to more effectively capture fine structural variations while maintaining perceptual integrity. Experiments on complex indoor and outdoor scenes demonstrate that the proposed method achieves superior detail preservation, reduced aliasing, and improved rendering quality compared with existing 3DGS approaches. These results highlight the potential of frequency- and depth-aware optimization for advancing computational and AI-driven imaging techniques.