Stereo Vision - Paper Copilot

Diffuse-Specular Separation using Binary Spherical GradientIllumination

Christos Kampouris; Stefanos Zafeiriou; Abhijeet Ghosh

Imperial College London

Portals

Abstract

We introduce a novel method for view-independent diffuse-specular separation of albedo and photometric normals without requiring polarization using binary spherical gradient illumination. The key idea is that with binary gradient illumination, a dielectric surface oriented towards the dark hemisphere exhibits pure diffuse reflectance while a surface oriented towards the bright hemisphere exhibits both diffuse and specular reflectance. We exploit this observation to formulate diffuse-specular separation based on color-space analysis of a surface’s response to binary spherical gradients and their complements. The method does not impose restrictions on viewpoints and requires fewer photographs for multiview acquisition than polarized spherical gradient illumination. We further demonstrate an efficient two-shot capture using spectral multiplexing of the illumination that enables diffuse-specular separation of albedo and heuristic separation of photometric normals.

Related Works

Photometric acquisition; Facial capture; Frequency domain separation; Binary spherical gradients

Extracting the Shape and Roughness of Specular Lobe Objects using Four Light Photometric Stereo

Fredric Solomon, Katsushi Ikeuchi

Carnegie Mellon University

Portals

Abstract

Two important aspects of part inspection are the measurement of surface shape and surface roughness. We propose a noncontact method of measuing surface shape and surface roughness. The method, which we call \"four light photometric stereo\" , uses four lights which sequentially illuminate the object under inspection, and a video camera for taking images of the object. Conceptually, the problem we are solving has three parts: shape extraction, pixel segmentation and roughness extraction. The shape information is produced directly by the three light and four light photometric stereo methods. After we have shape information, we can apply statistical segmentation techniques to determine which pixels are specular and which re non-specular. Then, we can use the specular pixels and shape information, in conjugation with a simplified Torrance-Sparrow reflectance model to determine the surface roughness The method has successfully been applied to a number of synthetic and real objects.

Circularly Polarized Spherical Illumination Reflectometry

Abhijeet Ghosh, Tongbo Chen, Pieter Peers, Cyrus A. Wilson, Paul Debevec

Portals

Abstract

We present a novel method for surface reflectometry from a few observations of a scene under a single uniform spherical field of circularly polarized illumination. The method is based on a novel analysis of the Stokes reflectance field of circularly polarized spherical illumination and yields per-pixel estimates of diffuse albedo, specular albedo, index of refraction, and specular roughness of isotropic BRDFs. To infer these reflectance parameters, we measure the Stokes parameters of the reflected light at each pixel by taking four photographs of the scene, consisting of three photographs with differently oriented linear polarizers in front of the camera, and one additional photograph with a circular polarizer. The method only assumes knowledge of surface orientation, for which we make a few additional photometric measurements. We verify our method with three different lighting setups, ranging from specialized to off-theshelf hardware, which project either discrete or continuous fields of spherical illumination. Our technique offers several benefits: it estimates a more detailed model of per-pixel surface reflectance parameters than previous work, it requires a relatively small number of measurements, it is applicable to a wide range of material types, and it is completely viewpoint independent.

Related Works

Reflectance Measurement and Model Fitting; Polarization-Based Reflectance Measurement; Reflectance Component Separation; Material Classification; Normal Estimation; Index of Refraction

Learning Deformable Tetrahedral Meshes for 3D Reconstruction

Jun Gao, Wenzheng Chen, Tommy Xiang, Clement Fuji Tsang, Alec Jacobson, Morgan McGuire, Sanja Fidler

NVIDIA; University of Toronto; Vector Institute

Portals

Abstract

3D shape representations that accommodate learning-based 3D reconstruction are an open problem in machine learning and computer graphics. Previous work on neural 3D reconstruction demonstrated benefits, but also limitations, of point cloud, voxel, surface mesh, and implicit function representations. We introduce Deformable Tetrahedral Meshes (DefTet) as a particular parameterization that utilizes volumetric tetrahedral meshes for the reconstruction problem. Unlike existing volumetric approaches, DefTet optimizes for both vertex placement and occupancy, and is differentiable with respect to standard 3D reconstruction loss functions. It is thus simultaneously high-precision, volumetric, and amenable to learning-based neural architectures. We show that it can represent arbitrary, complex topology, is both memory and computationally efficient, and can produce high-fidelity reconstructions with a significantly smaller grid size than alternative volumetric approaches. The predicted surfaces are also inherently defined as tetrahedral meshes, thus do not require post-processing. We demonstrate that DefTet matches or exceeds both the quality of the previous best approaches and the performance of the fastest ones. Our approach obtains high-quality tetrahedral meshes computed directly from noisy point clouds, and is the first to showcase high-quality 3D tet-mesh results using only a single image as input.

Related Works

3D Reconstruction; Tetrahedral Meshing

Comparisons

3D-R2N2, DeepMCube, Pixel2mesh, OccNet, MeshRCNN

Acquiring Reflectance and Shape from Continuous Spherical Harmonic Illumination

Borom Tunwattanapong, Graham Fyffe, Paul Graham, Jay Busch, Xueming Yu, Abhijeet Ghosh, Paul Debevec

USC Institute for Creative Technologies; Imperial College London

Portals

Abstract

We present a novel technique for acquiring the geometry and spatially-varying reflectance properties of 3D objects by observing them under continuous spherical harmonic illumination conditions. The technique is general enough to characterize either entirely specular or entirely diffuse materials, or any varying combination across the surface of the object. We employ a novel computational illumination setup consisting of a rotating arc of controllable LEDs which sweep out programmable spheres of incident illumination during 1-second exposures. We illuminate the object with a succession of spherical harmonic illumination conditions, as well as photographed environmental lighting for validation. From the response of the object to the harmonics, we can separate diffuse and specular reflections, estimate world-space diffuse and specular normals, and compute anisotropic roughness parameters for each view of the object. We then use the maps of both diffuse and specular reflectance to form correspondences in a multiview stereo algorithm, which allows even highly specular surfaces to be corresponded across views. The algorithm yields a complete 3D model and a set of merged reflectance maps. We use this technique to digitize the shape and reflectance of a variety of objects difficult to acquire with other techniques and present validation renderings which match well to photographs in similar lighting.

Related Works

Spatially Varying BRDF Capture; Using Extended Light Sources; Reflectance from Spherical Illumination; Geometry from Specular Reflection

Measuring and Modeling Anisotropic Reflection

Greg Ward

Lawrence Berkeley Laboratory

Portals

Abstract

A new device for measuring the spatial reflectancedistributionsof surfaces is introduced, along with a new mathematical model of sniaorropic reflectance. The reflectance model presented is both simple and accurate, permitting efficient reflectance data reduction rasdreproduction. Tire validity of the model is substantiated with comparisons to complete meaarsremems of surface reflectance functions gathered with the novel retlectometry device. This new device uses imaging technology to capture the entire hemisphem of reflected directions simttkarreously, which greatly accelerates the reflectance data gathering process, making it pssible to measure dozens of surfaces in the time that it used to take to do one. Example measurements and simulations are shown. and a table of fitted parameters for several surfaces is presented.

DeRF: Decomposed Radiance Fields

Daniel Rebain, Wei Jiang, Soroosh Yazdani, Ke Li, Kwang Moo Yi, Andrea Tagliasacchi

University of British Columbia; Simon Fraser University; University of Toronto; Google Research

Portals

Abstract

With the advent of Neural Radiance Fields (NeRF), neural networks can now render novel views of a 3D scene with quality that fools the human eye. Yet, generating these images is very computationally intensive, limiting their applicability in practical scenarios. In this paper, we propose a technique based on spatial decomposition capable of mitigating this issue. Our key observation is that there are diminishing returns in employing larger (deeper and/or wider) networks. Hence, we propose to spatially decompose a scene and dedicate smaller networks for each decomposed part. When working together, these networks can render the whole scene. This allows us near-constant inference time regardless of the number of decomposed parts. Moreover, we show that a Voronoi spatial decomposition is preferable for this purpose, as it is provably compatible with the Painter's Algorithm for efficient and GPU-friendly rendering. Our experiments show that for real-world scenes, our method provides up to 3x more efficient inference than NeRF (with the same rendering quality), or an improvement of up to 1.0~dB in PSNR (for the same inference cost).

Related Works

Image-space neural rendering; Neural volumetric rendering

Comparisons

NeRF

NSVF: Neural Sparse Voxel Fields

Error: Cannot create object

Category: Stereo Vision

Portals

Abstract

Related Works

Portals

Abstract

Portals

Abstract

Related Works

Portals

Abstract

Related Works

Comparisons

Portals

Abstract

Related Works

Portals

Abstract

Portals

Abstract

Related Works

Comparisons