Global ETD Search
Search theses and dissertations gathered from participating repositories worldwide. Every result links back to the library that holds it. No account is needed.
Results
Showing 1 to 20 of 39 for “"Radiance Field"”.
-
Illumination-Aware 3D Gaussian Splatting for Spacecraft Reconstruction and Scene Composition
Radiance fields have recently emerged as a powerful tool for learning continuous, three-dimensional representations of complex scenes from collections of discrete image observations. Neural Radiance Fields and 3D Gaussian Splatting have demonstrated high-quality novel view synthesis across a wide …
-
Geometry-aided 3D image processing
… (a representation for low-frequency 3D-direction fields) to model the illumination for the relighting task.Last, we explore how geometry shapes could help with a deep learning algorithm for a particular field of both locations and directions, the dynamical neural radiance field for human heads. …
-
Advancing Photometric Odometry to Dense Volumetric Simultaneous Localization and Mapping
… the visual SLAM domain. Techniques like Neural Radiance Fields (NeRF) and 3D Gaussian Splatting (3DGS) enable the rapid generation of dense volumetric scenes from RGB images. Researchers have integrated these radiance field techniques into SLAM to address a key limitation of traditional systems. …
-
Radiance in the ocean: effects of wave slope and raman scattering near the surface and at depths through the asymptotic region
… were conducted on the nature of the radiance field in clear ocean water. It is important to understand the sunlight intensity below the sea surface because this leads to an understanding of how ocean creatures navigate in shallow and deep water. The nature of the radiance field is …
-
Instant HDR-NeRF: Fast Learning Of High Dynamic Range View Synthesis With Unknown Exposure Settings
We propose Instant High Dynamic Range Neural Radiance Fields (Instant HDR-NeRF), a method of learning high dynamic range (HDR) view synthesis from a set of low dynamic range (LDR) views with unknown and varying exposure and white balance in as little as minutes. Our method can render novel HDR …
-
Robust 3D content generation with acquisition aware frame selection for Gaussian Splatting
… a wide range of applications across different fields, including entertainment, cultural preservation, virtual or augmented reality education environments, and medical training spaces. Traditional methods for creating 3D content are often time-consuming, require specialized expertise, and can be …
-
Robust 3D content generation with acquisition aware frame selection for Gaussian Splatting
… a wide range of applications across different fields, including entertainment, cultural preservation, virtual or augmented reality education environments, and medical training spaces. Traditional methods for creating 3D content are often time-consuming, require specialized expertise, and can be …
-
Geometry-Aware Diffusion Models for Multiview Scene Inpainting
… by combining generative models with a 3D radiance field to fuse information across a relatively dense set of viewpoints. However, a major drawback of these methods is that they often produce blurry images due to the fusion of inconsistent cross-view images. To avoid blurry inpaintings, we …
-
View Enhancement of Arthroscopic Surgery Through Augmented Reality and Neural Radiance Fields
… explore a novel framework that combines Neural Radiance Field (NeRF) based reconstruction and Augmented Reality (AR) to overcome challenges and innovations in arthroscopy, a minimally invasive orthopedic procedure that is commonly used for the treatment of joints such as the shoulder, hip, and …
-
Patch-Based Material Transfer for Mesh-Mapped 3D Gaussian Splatting
… 3D Gaussian Splatting (3DGS) have revolutionized radiance field reconstruction, yet these models remain static and tied to their captured geometry. There is currently no straightforward method to apply 3DGS surface details to custom objects. This thesis introduces a novel material transfer …
-
4D-VIDEO DYNAMIC SCENE RECONSTRUCTION AND EDITING
… we propose DeVRF, a novel fast deformable voxel radiance field to achieve efficient reconstruction of dynamic scenes within 10 minutes, which is 100 times faster than SOTA with high fidelity. Secondly, the thesis introduces HOSNeRF, a novel free-viewpoint rendering method that reconstructs neural …
-
DeepPanoRF: Deep Prior for Neural 3D Reconstruction From Sparse Panoramas
Advances in neural field representations have led to a significant improvement in view synthesis quality. However, many current novel view synthesis methods rely on a dense set of input views, which can be impractical and inefficient in real-world applications. We propose DeepPanoRF, a novel method …
-
Multiple Scattering Contributions to Atmospheric Radiance with Applications to Remote Sensing
… as a function of wavelength to measured sky path radiances. The purpose of doing this is to estimate what percentage of measured radiance is due to multiple scattering, provide data for possible correction algorithms that attempt to account for multiple scattering, and provide needed data that …
-
Uncertainty-aware lidar-visual 3d reconstruction
… advances in learning-based depth estimation and radiance fields offer promising alternatives to classical reconstruction methods, enabling new possibilities for reality capture, robot simulation, and navigation. However, applying these techniques to large-scale outdoor 3D reconstruction remains …
-
Vector radiative transfer and its application to the remote sensing of aerosols and hydrosols
… model. It is now able to simulate the vector radiance field in a plane parallel system of a coupled atmosphere and ocean with a wind roughened interface. This allows to more accurately simulate the radiance field, since the simplifications of a scalar model can cause errors of the simulated …
-
Active Perception for 3D Scene Representations in Robotics from an Information Theoretic Perspective
… of active robot perception systems utilizing radiance field representations, particularly 3D Gaussian Splatting, from an information-theoretic perspective. We first address the challenge of quantifying observed information in 3D scene representations. By leveraging the Fisher Information …
-
High-Accuracy Intraoperative 3D Reconstruction with Neural Rendering Technology
… neural rendering techniques—principally Neural Radiance Fields (NeRF) and 3D Gaussian Splatting—can be redesigned from photorealistic view-synthesis tools into quantitative metrology instruments, recovering near-CT geometric accuracy over the surgically exposed field directly from routine …
-
Towards High Fidelity Generation of Synthetic 3D Worlds
… by articles we make several contributions to the field of deep learning, specifically in the areas of generative modelling and 3D scene synthesis. Each article develops an important component in the full training pipeline: evaluation metrics, constructing datasets, and model architecture. In our …
Page 1 of 2