Showing papers on "3D reconstruction published in 2018"

Occupancy Networks: Learning 3D Reconstruction in Function Space

Abstract: With the advent of deep neural networks, learning-based approaches for 3D reconstruction have gained popularity. However, unlike for images, in 3D there is no canonical representation which is both computationally and memory efficient yet allows for representing high-resolution geometry of arbitrary topology. Many of the state-of-the-art learning-based 3D reconstruction approaches can hence only represent very coarse 3D geometry or are limited to a restricted domain. In this paper, we propose Occupancy Networks, a new representation for learning-based 3D reconstruction methods. Occupancy networks implicitly represent the 3D surface as the continuous decision boundary of a deep neural network classifier. In contrast to existing approaches, our representation encodes a description of the 3D output at infinite resolution without excessive memory footprint. We validate that our representation can efficiently encode 3D structure and can be inferred from various kinds of input. Our experiments demonstrate competitive results, both qualitatively and quantitatively, for the challenging tasks of 3D reconstruction from single images, noisy point clouds and coarse discrete voxel grids. We believe that occupancy networks will become a useful tool in a wide variety of learning-based 3D tasks.

DiffuserCam: lensless single-exposure 3D imaging

Abstract: We demonstrate a compact, easy-to-build computational camera for single-shot three-dimensional (3D) imaging. Our lensless system consists solely of a diffuser placed in front of an image sensor. Every point within the volumetric field-of-view projects a unique pseudorandom pattern of caustics on the sensor. By using a physical approximation and simple calibration scheme, we solve the large-scale inverse problem in a computationally efficient way. The caustic patterns enable compressed sensing, which exploits sparsity in the sample to solve for more 3D voxels than pixels on the 2D sensor. Our 3D reconstruction grid is chosen to match the experimentally measured two-point optical resolution, resulting in 100 million voxels being reconstructed from a single 1.3 megapixel image. However, the effective resolution varies significantly with scene content. Because this effect is common to a wide range of computational cameras, we provide a new theory for analyzing resolution in such systems.

Deep Virtual Stereo Odometry: Leveraging Deep Depth Prediction for Monocular Direct Sparse Odometry

Abstract: Monocular visual odometry approaches that purely rely on geometric cues are prone to scale drift and require sufficient motion parallax in successive frames for motion estimation and 3D reconstruction. In this paper, we propose to leverage deep monocular depth prediction to overcome limitations of geometry-based monocular visual odometry. To this end, we incorporate deep depth predictions into Direct Sparse Odometry (DSO) as direct virtual stereo measurements. For depth prediction, we design a novel deep network that refines predicted depth from a single image in a two-stage process. We train our network in a semi-supervised way on photoconsistency in stereo images and on consistency with accurate sparse depth reconstructions from Stereo DSO. Our deep predictions excel state-of-the-art approaches for monocular depth on the KITTI benchmark. Moreover, our Deep Virtual Stereo Odometry clearly exceeds previous monocular and deep-learning based methods in accuracy. It even achieves comparable performance to the state-of-the-art stereo methods, while only relying on a single camera.

Pix3D: Dataset and Methods for Single-Image 3D Shape Modeling

Abstract: We study 3D shape modeling from a single image and make contributions to it in three aspects. First, we present Pix3D, a large-scale benchmark of diverse image-shape pairs with pixel-level 2D-3D alignment. Pix3D has wide applications in shape-related tasks including reconstruction, retrieval, viewpoint estimation, etc. Building such a large-scale dataset, however, is highly challenging; existing datasets either contain only synthetic data, or lack precise alignment between 2D images and 3D shapes, or only have a small number of images. Second, we calibrate the evaluation criteria for 3D shape reconstruction through behavioral studies, and use them to objectively and systematically benchmark cutting-edge reconstruction algorithms on Pix3D. Third, we design a novel model that simultaneously performs 3D reconstruction and pose estimation; our multi-task learning approach achieves state-of-the-art performance on both tasks.

Visual SLAM and Structure from Motion in Dynamic Environments: A Survey

Abstract: In the last few decades, Structure from Motion (SfM) and visual Simultaneous Localization and Mapping (visual SLAM) techniques have gained significant interest from both the computer vision and robotic communities. Many variants of these techniques have started to make an impact in a wide range of applications, including robot navigation and augmented reality. However, despite some remarkable results in these areas, most SfM and visual SLAM techniques operate based on the assumption that the observed environment is static. However, when faced with moving objects, overall system accuracy can be jeopardized. In this article, we present for the first time a survey of visual SLAM and SfM techniques that are targeted toward operation in dynamic environments. We identify three main problems: how to perform reconstruction (robust visual SLAM), how to segment and track dynamic objects, and how to achieve joint motion segmentation and reconstruction. Based on this categorization, we provide a comprehensive taxonomy of existing approaches. Finally, the advantages and disadvantages of each solution class are critically discussed from the perspective of practicality and robustness.

Hand Pose Estimation via Latent 2.5D Heatmap Regression

Abstract: Estimating the 3D pose of a hand is an essential part of human-computer interaction. Estimating 3D pose using depth or multi-view sensors has become easier with recent advances in computer vision, however, regressing pose from a single RGB image is much less straightforward. The main difficulty arises from the fact that 3D pose requires some form of depth estimates, which are ambiguous given only an RGB image. In this paper we propose a new method for 3D hand pose estimation from a monocular image through a novel 2.5D pose representation. Our new representation estimates pose up to a scaling factor, which can be estimated additionally if a prior of the hand size is given. We implicitly learn depth maps and heatmap distributions with a novel CNN architecture. Our system achieves state-of-the-art accuracy for 2D and 3D hand pose estimation on several challenging datasets in presence of severe occlusions.

Nonlinear 3D Face Morphable Model

Abstract: As a classic statistical model of 3D facial shape and texture, 3D Morphable Model (3DMM) is widely used in facial analysis, e.g., model fitting, image synthesis. Conventional 3DMM is learned from a set of well-controlled 2D face images with associated 3D face scans, and represented by two sets of PCA basis functions. Due to the type and amount of training data, as well as the linear bases, the representation power of 3DMM can be limited. To address these problems, this paper proposes an innovative framework to learn a nonlinear 3DMM model from a large set of unconstrained face images, without collecting 3D face scans. Specifically, given a face image as input, a network encoder estimates the projection, shape and texture parameters. Two decoders serve as the nonlinear 3DMM to map from the shape and texture parameters to the 3D shape and texture, respectively. With the projection parameter, 3D shape, and texture, a novel analytically-differentiable rendering layer is designed to reconstruct the original input face. The entire network is end-to-end trainable with only weak supervision. We demonstrate the superior representation power of our nonlinear 3DMM over its linear counterpart, and its contribution to face alignment and 3D reconstruction.

3D-SIS: 3D Semantic Instance Segmentation of RGB-D Scans

Abstract: We introduce 3D-SIS, a novel neural network architecture for 3D semantic instance segmentation in commodity RGB-D scans. The core idea of our method is to jointly learn from both geometric and color signal, thus enabling accurate instance predictions. Rather than operate solely on 2D frames, we observe that most computer vision applications have multi-view RGB-D input available, which we leverage to construct an approach for 3D instance segmentation that effectively fuses together these multi-modal inputs. Our network leverages high-resolution RGB input by associating 2D images with the volumetric grid based on the pose alignment of the 3D reconstruction. For each image, we first extract 2D features for each pixel with a series of 2D convolutions; we then backproject the resulting feature vector to the associated voxel in the 3D grid. This combination of 2D and 3D feature learning allows significantly higher accuracy object detection and instance segmentation than state-of-the-art alternatives. We show results on both synthetic and real-world public benchmarks, achieving an improvement in mAP of over 13 on real-world data.

Nonlinear 3D Face Morphable Model

Abstract: As a classic statistical model of 3D facial shape and texture, 3D Morphable Model (3DMM) is widely used in facial analysis, e.g., model fitting, image synthesis. Conventional 3DMM is learned from a set of well-controlled 2D face images with associated 3D face scans, and represented by two sets of PCA basis functions. Due to the type and amount of training data, as well as the linear bases, the representation power of 3DMM can be limited. To address these problems, this paper proposes an innovative framework to learn a nonlinear 3DMM model from a large set of unconstrained face images, without collecting 3D face scans. Specifically, given a face image as input, a network encoder estimates the projection, shape and texture parameters. Two decoders serve as the nonlinear 3DMM to map from the shape and texture parameters to the 3D shape and texture, respectively. With the projection parameter, 3D shape, and texture, a novel analytically-differentiable rendering layer is designed to reconstruct the original input face. The entire network is end-to-end trainable with only weak supervision. We demonstrate the superior representation power of our nonlinear 3DMM over its linear counterpart, and its contribution to face alignment and 3D reconstruction.

EMVS: Event-Based Multi-View Stereo—3D Reconstruction with an Event Camera in Real-Time

Abstract: Event cameras are bio-inspired vision sensors that output pixel-level brightness changes instead of standard intensity frames. They offer significant advantages over standard cameras, namely a very high dynamic range, no motion blur, and a latency in the order of microseconds. However, because the output is composed of a sequence of asynchronous events rather than actual intensity images, traditional vision algorithms cannot be applied, so that a paradigm shift is needed. We introduce the problem of event-based multi-view stereo (EMVS) for event cameras and propose a solution to it. Unlike traditional MVS methods, which address the problem of estimating dense 3D structure from a set of known viewpoints, EMVS estimates semi-dense 3D structure from an event camera with known trajectory. Our EMVS solution elegantly exploits two inherent properties of an event camera: (1) its ability to respond to scene edges—which naturally provide semi-dense geometric information without any pre-processing operation—and (2) the fact that it provides continuous measurements as the sensor moves. Despite its simplicity (it can be implemented in a few lines of code), our algorithm is able to produce accurate, semi-dense depth maps, without requiring any explicit data association or intensity estimation. We successfully validate our method on both synthetic and real data. Our method is computationally very efficient and runs in real-time on a CPU.

Pix3D: Dataset and Methods for Single-Image 3D Shape Modeling.

Abstract: We study 3D shape modeling from a single image and make contributions to it in three aspects. First, we present Pix3D, a large-scale benchmark of diverse image-shape pairs with pixel-level 2D-3D alignment. Pix3D has wide applications in shape-related tasks including reconstruction, retrieval, viewpoint estimation, etc. Building such a large-scale dataset, however, is highly challenging; existing datasets either contain only synthetic data, or lack precise alignment between 2D images and 3D shapes, or only have a small number of images. Second, we calibrate the evaluation criteria for 3D shape reconstruction through behavioral studies, and use them to objectively and systematically benchmark cutting-edge reconstruction algorithms on Pix3D. Third, we design a novel model that simultaneously performs 3D reconstruction and pose estimation; our multi-task learning approach achieves state-of-the-art performance on both tasks.

A comparison of volumetric information gain metrics for active 3D object reconstruction

Abstract: In this paper, we investigate the following question: when performing next best view selection for volumetric 3D reconstruction of an object by a mobile robot equipped with a dense (camera-based) depth sensor, what formulation of information gain is best? To address this question, we propose several new ways to quantify the volumetric information (VI) contained in the voxels of a probabilistic volumetric map, and compare them to the state of the art with extensive simulated experiments. Our proposed formulations incorporate factors such as visibility likelihood and the likelihood of seeing new parts of the object. The results of our experiments allow us to draw some clear conclusions about the VI formulations that are most effective in different mobile-robot reconstruction scenarios. To the best of our knowledge, this is the first comparative survey of VI formulation performance for active 3D object reconstruction. Additionally, our modular software framework is adaptable to other robotic platforms and general reconstruction problems, and we release it open source for autonomous reconstruction tasks.

DeformNet: Free-Form Deformation Network for 3D Shape Reconstruction from a Single Image

Abstract: 3D reconstruction from a single image is a key problem in multiple applications ranging from robotic manipulation to augmented reality. Prior methods have tackled this problem through generative models which predict 3D reconstructions as voxels or point clouds. However, these methods can be computationally expensive and miss fine details. We introduce a new differentiable layer for 3D data deformation and use it in DEFORMNET to learn a model for 3D reconstruction-through-deformation. DEFORMNET takes an image input, finds a nearest shape template from a database, and deforms the template to match the query image. We evaluate our approach on the ShapeNet dataset and show that - (a) the Free-Form Deformation layer is a powerful new building block for Deep Learning models that manipulate 3D data (b) DEFORMNET uses this FFD layer combined with shape retrieval for smooth and detail-preserving 3D reconstruction of qualitatively plausible point clouds with respect to a single query image (c) compared to other state-of-the-art 3D reconstruction methods, DEFORMNET quantitatively matches or outperforms their benchmarks by significant margins.

Learning Shape Priors for Single-View 3D Completion and Reconstruction

Abstract: The problem of single-view 3D shape completion or reconstruction is challenging, because among the many possible shapes that explain an observation, most are implausible and do not correspond to natural objects Recent research in the field has tackled this problem by exploiting the expressiveness of deep convolutional networks In fact, there is another level of ambiguity that is often overlooked: among plausible shapes, there are still multiple shapes that fit the 2D image equally well; ie, the ground truth shape is non-deterministic given a single-view input Existing fully supervised approaches fail to address this issue, and often produce blurry mean shapes with smooth surfaces but no fine details In this paper, we propose ShapeHD, pushing the limit of single-view shape completion and reconstruction by integrating deep generative models with adversarially learned shape priors The learned priors serve as a regularizer, penalizing the model only if its output is unrealistic, not if it deviates from the ground truth Our design thus overcomes both levels of ambiguity aforementioned Experiments demonstrate that ShapeHD outperforms state of the art by a large margin in both shape completion and shape reconstruction on multiple real datasets

Semi-Dense 3D Reconstruction with a Stereo Event Camera

Abstract: Event cameras are bio-inspired sensors that offer several advantages, such as low latency, high-speed and high dynamic range, to tackle challenging scenarios in computer vision. This paper presents a solution to the problem of 3D reconstruction from data captured by a stereo event-camera rig moving in a static scene, such as in the context of stereo Simultaneous Localization and Mapping. The proposed method consists of the optimization of an energy function designed to exploit small-baseline spatio-temporal consistency of events triggered across both stereo image planes. To improve the density of the reconstruction and to reduce the uncertainty of the estimation, a probabilistic depth-fusion strategy is also developed. The resulting method has no special requirements on either the motion of the stereo event-camera rig or on prior knowledge about the scene. Experiments demonstrate our method can deal with both texture-rich scenes as well as sparse scenes, outperforming state-of-the-art stereo methods based on event data image representations.

3D-LMNet: Latent Embedding Matching for Accurate and Diverse 3D Point Cloud Reconstruction from a Single Image

Abstract: 3D reconstruction from single view images is an ill-posed problem. Inferring the hidden regions from self-occluded images is both challenging and ambiguous. We propose a two-pronged approach to address these issues. To better incorporate the data prior and generate meaningful reconstructions, we propose 3D-LMNet, a latent embedding matching approach for 3D reconstruction. We first train a 3D point cloud auto-encoder and then learn a mapping from the 2D image to the corresponding learnt embedding. To tackle the issue of uncertainty in the reconstruction, we predict multiple reconstructions that are consistent with the input view. This is achieved by learning a probablistic latent space with a novel view-specific diversity loss. Thorough quantitative and qualitative analysis is performed to highlight the significance of the proposed approach. We outperform state-of-the-art approaches on the task of single-view 3D reconstruction on both real and synthetic datasets while generating multiple plausible reconstructions, demonstrating the generalizability and utility of our approach.

Practical SVBRDF acquisition of 3D objects with unstructured flash photography

Abstract: Capturing spatially-varying bidirectional reflectance distribution functions (SVBRDFs) of 3D objects with just a single, hand-held camera (such as an off-the-shelf smartphone or a DSLR camera) is a difficult, open problem. Previous works are either limited to planar geometry, or rely on previously scanned 3D geometry, thus limiting their practicality. There are several technical challenges that need to be overcome: First, the built-in flash of a camera is almost colocated with the lens, and at a fixed position; this severely hampers sampling procedures in the light-view space. Moreover, the near-field flash lights the object partially and unevenly. In terms of geometry, existing multiview stereo techniques assume diffuse reflectance only, which leads to overly smoothed 3D reconstructions, as we show in this paper. We present a simple yet powerful framework that removes the need for expensive, dedicated hardware, enabling practical acquisition of SVBRDF information from real-world, 3D objects with a single, off-the-shelf camera with a built-in flash. In addition, by removing the diffuse reflection assumption and leveraging instead such SVBRDF information, our method outputs high-quality 3D geometry reconstructions, including more accurate high-frequency details than state-of-the-art multiview stereo techniques. We formulate the joint reconstruction of SVBRDFs, shading normals, and 3D geometry as a multi-stage, iterative inverse-rendering reconstruction pipeline. Our method is also directly applicable to any existing multiview 3D reconstruction technique. We present results of captured objects with complex geometry and reflectance; we also validate our method numerically against other existing approaches that rely on dedicated hardware, additional sources of information, or both.

3D Human Body Reconstruction from a Single Image via Volumetric Regression

Abstract: This paper proposes the use of an end-to-end Convolutional Neural Network for direct reconstruction of the 3D geometry of humans via volumetric regression The proposed method does not require the fitting of a shape model and can be trained to work from a variety of input types, whether it be landmarks, images or segmentation masks Additionally, non-visible parts, either self-occluded or otherwise, are still reconstructed, which is not the case with depth map regression We present results that show that our method can handle both pose variation and detailed reconstruction given appropriate datasets for training

Plan3D: Viewpoint and Trajectory Optimization for Aerial Multi-View Stereo Reconstruction

Abstract: We introduce a new method that efficiently computes a set of viewpoints and trajectories for high-quality 3D reconstructions in outdoor environments. Our goal is to automatically explore an unknown area and obtain a complete 3D scan of a region of interest (e.g., a large building). Images from a commodity RGB camera, mounted on an autonomously navigated quadcopter, are fed into a multi-view stereo reconstruction pipeline that produces high-quality results but is computationally expensive. In this setting, the scanning result is constrained by the restricted flight time of quadcopters. To this end, we introduce a novel optimization strategy that respects these constraints by maximizing the information gain from sparsely sampled viewpoints while limiting the total travel distance of the quadcopter. At the core of our method lies a hierarchical volumetric representation that allows the algorithm to distinguish between unknown, free, and occupied space. Furthermore, our information gain-based formulation leverages this representation to handle occlusions in an efficient manner. In addition to the surface geometry, we utilize free-space information to avoid obstacles and determine collision-free flight paths. Our tool can be used to specify the region of interest and to plan trajectories. We demonstrate our method by obtaining a number of compelling 3D reconstructions, and we provide a thorough quantitative evaluation showing improvement over previous state-of-the-art and regular patterns.

Image2Mesh: A learning framework for single image 3D reconstruction

Abstract: A challenge that remains open in 3D deep learning is how to efficiently represent 3D data to feed deep neural networks. Recent works have been relying on volumetric or point cloud representations, but such approaches suffer from a number of issues such as computational complexity, unordered data, and lack of finer geometry. An efficient way to represent a 3D shape is through a polygon mesh as it encodes both shape’s geometric and topological information. However, the mesh’s data structure is an irregular graph (i.e. collection of vertices connected by edges to form polygonal faces) and it is not straightforward to integrate it into learning frameworks since every mesh is likely to have a different structure. Here we address this drawback by efficiently converting an unstructured 3D mesh into a regular and compact shape parametrization that is ready for machine learning applications. We developed a simple and lightweight learning framework able to reconstruct high-quality 3D meshes from a single image by using a compact representation that encodes a mesh using free-form deformation and sparse linear combination in a small dictionary of 3D models. In contrast to prior work, we do not rely on classical silhouette and landmark registration techniques to perform the 3D reconstruction. We extensively evaluated our method on synthetic and real-world datasets and found that it can efficiently and compactly reconstruct 3D objects while preserving its important geometrical aspects.

Normal Integration: A Survey

Abstract: The need for efficient normal integration methods is driven by several computer vision tasks such as shape-from-shading, photometric stereo, deflectometry. In the first part of this survey, we select the most important properties that one may expect from a normal integration method, based on a thorough study of two pioneering works by Horn and rooks (Comput Vis Graph Image Process 33(2): 174–208, 1986) and Frankot and Chellappa (IEEE Trans Pattern Anal Mach Intell 10(4): 439-451, 1988). Apart from accuracy, an integration method should at least be fast and robust to a noisy normal field. In addition, it should be able to handle several types of boundary condition, including the case of a free boundary and a reconstruction domain of any shape, i.e., which is not necessarily rectangular. It is also much appreciated that a minimum number of parameters have to be tuned, or even no parameter at all. Finally, it should preserve the depth discontinuities. In the second part of this survey, we review most of the existing methods in view of this analysis and conclude that none of them satisfies all of the required properties. This work is complemented by a companion paper entitled Variational Methods for Normal Integration, in which we focus on the problem of normal integration in the presence of depth discontinuities, a problem which occurs as soon as there are occlusions.

Texture Mapping for 3D Reconstruction with RGB-D Sensor

Abstract: Acquiring realistic texture details for 3D models is important in 3D reconstruction. However, the existence of geometric errors, caused by noisy RGB-D sensor data, always makes the color images cannot be accurately aligned onto reconstructed 3D models. In this paper, we propose a global-to-local correction strategy to obtain more desired texture mapping results. Our algorithm first adaptively selects an optimal image for each face of the 3D model, which can effectively remove blurring and ghost artifacts produced by multiple image blending. We then adopt a non-rigid global-to-local correction step to reduce the seaming effect between textures. This can effectively compensate for the texture and the geometric misalignment caused by camera pose drift and geometric errors. We evaluate the proposed algorithm in a range of complex scenes and demonstrate its effective performance in generating seamless high fidelity textures for 3D models.

Learning to Reconstruct Shapes from Unseen Classes

Abstract: From a single image, humans are able to perceive the full 3D shape of an object by exploiting learned shape priors from everyday life. Contemporary single-image 3D reconstruction algorithms aim to solve this task in a similar fashion, but often end up with priors that are highly biased by training classes. Here we present an algorithm, Generalizable Reconstruction (GenRe), designed to capture more generic, class-agnostic shape priors. We achieve this with an inference network and training procedure that combine 2.5D representations of visible surfaces (depth and silhouette), spherical shape representations of both visible and non-visible surfaces, and 3D voxel-based representations, in a principled manner that exploits the causal structure of how 3D shapes give rise to 2D images. Experiments demonstrate that GenRe performs well on single-view shape reconstruction, and generalizes to diverse novel objects from categories not seen during training.

Road Surface 3D Reconstruction Based on Dense Subpixel Disparity Map Estimation

Abstract: Various 3D reconstruction methods have enabled civil engineers to detect damage on a road surface. To achieve the millimeter accuracy required for road condition assessment, a disparity map with subpixel resolution needs to be used. However, none of the existing stereo matching algorithms are specially suitable for the reconstruction of the road surface. Hence in this paper, we propose a novel dense subpixel disparity estimation algorithm with high computational efficiency and robustness. This is achieved by first transforming the perspective view of the target frame into the reference view, which not only increases the accuracy of the block matching for the road surface but also improves the processing speed. The disparities are then estimated iteratively using our previously published algorithm, where the search range is propagated from three estimated neighboring disparities. Since the search range is obtained from the previous iteration, errors may occur when the propagated search range is not sufficient. Therefore, a correlation maxima verification is performed to rectify this issue, and the subpixel resolution is achieved by conducting a parabola interpolation enhancement. Furthermore, a novel disparity global refinement approach developed from the Markov random fields and fast bilateral stereo is introduced to further improve the accuracy of the estimated disparity map, where disparities are updated iteratively by minimizing the energy function that is related to their interpolated correlation polynomials. The algorithm is implemented in C language with a near real-time performance. The experimental results illustrate that the absolute error of the reconstruction varies from 0.1 to 3 mm.

Learning to Reconstruct Shapes from Unseen Classes

Abstract: From a single image, humans are able to perceive the full 3D shape of an object by exploiting learned shape priors from everyday life. Contemporary single-image 3D reconstruction algorithms aim to solve this task in a similar fashion, but often end up with priors that are highly biased by training classes. Here we present an algorithm, Generalizable Reconstruction (GenRe), designed to capture more generic, class-agnostic shape priors. We achieve this with an inference network and training procedure that combine 2.5D representations of visible surfaces (depth and silhouette), spherical shape representations of both visible and non-visible surfaces, and 3D voxel-based representations, in a principled manner that exploits the causal structure of how 3D shapes give rise to 2D images. Experiments demonstrate that GenRe performs well on single-view shape reconstruction, and generalizes to diverse novel objects from categories not seen during training.

Road surface 3d reconstruction based on dense subpixel disparity map estimation

Abstract: Various 3D reconstruction methods have enabled civil engineers to detect damage on a road surface. To achieve the millimetre accuracy required for road condition assessment, a disparity map with subpixel resolution needs to be used. However, none of the existing stereo matching algorithms are specially suitable for the reconstruction of the road surface. Hence in this paper, we propose a novel dense subpixel disparity estimation algorithm with high computational efficiency and robustness. This is achieved by first transforming the perspective view of the target frame into the reference view, which not only increases the accuracy of the block matching for the road surface but also improves the processing speed. The disparities are then estimated iteratively using our previously published algorithm where the search range is propagated from three estimated neighbouring disparities. Since the search range is obtained from the previous iteration, errors may occur when the propagated search range is not sufficient. Therefore, a correlation maxima verification is performed to rectify this issue, and the subpixel resolution is achieved by conducting a parabola interpolation enhancement. Furthermore, a novel disparity global refinement approach developed from the Markov Random Fields and Fast Bilateral Stereo is introduced to further improve the accuracy of the estimated disparity map, where disparities are updated iteratively by minimising the energy function that is related to their interpolated correlation polynomials. The algorithm is implemented in C language with a near real-time performance. The experimental results illustrate that the absolute error of the reconstruction varies from 0.1 mm to 3 mm.

Fast depth densification for occlusion-aware augmented reality

Abstract: Current AR systems only track sparse geometric features but do not compute depth for all pixels. For this reason, most AR effects are pure overlays that can never be occluded by real objects. We present a novel algorithm that propagates sparse depth to every pixel in near realtime. The produced depth maps are spatio-temporally smooth but exhibit sharp discontinuities at depth edges. This enables AR effects that can fully interact with and be occluded by the real scene. Our algorithm uses a video and a sparse SLAM reconstruction as input. It starts by estimating soft depth edges from the gradient of optical flow fields. Because optical flow is unreliable near occlusions we compute forward and backward flow fields and fuse the resulting depth edges using a novel reliability measure. We then localize the depth edges by thinning and aligning them with image edges. Finally, we optimize the propagated depth smoothly but encourage discontinuities at the recovered depth edges. We present results for numerous real-world examples and demonstrate the effectiveness for several occlusion-aware AR video effects. To quantitatively evaluate our algorithm we characterize the properties that make depth maps desirable for AR applications, and present novel evaluation metrics that capture how well these are satisfied. Our results compare favorably to a set of competitive baseline algorithms in this context.

DeepCalib: a deep learning approach for automatic intrinsic calibration of wide field-of-view cameras

Abstract: Calibration of wide field-of-view cameras is a fundamental step for numerous visual media production applications, such as 3D reconstruction, image undistortion, augmented reality and camera motion estimation. However, existing calibration methods require multiple images of a calibration pattern (typically a checkerboard), assume the presence of lines, require manual interaction and/or need an image sequence. In contrast, we present a novel fully automatic deep learning-based approach that overcomes all these limitations and works with a single image of general scenes. Our approach builds upon the recent developments in deep Convolutional Neural Networks (CNN): our network automatically estimates the intrinsic parameters of the camera (focal length and distortion parameter) from a single input image. In order to train the CNN, we leverage the great amount of omnidirectional images available on the Internet to automatically generate a large-scale dataset composed of millions of wide field-of-view images with ground truth intrinsic parameters. Experiments successfully demonstrated the quality of our results, both quantitatively and qualitatively.