Motion Segmentation of Truncated Signed Distance Function Based Volumetric Surfaces

@article{Perera2015MotionSO,
  title={Motion Segmentation of Truncated Signed Distance Function Based Volumetric Surfaces},
  author={Samunda Perera and Nick Barnes and Xuming He and Shahram Izadi and Pushmeet Kohli and Ben Glocker},
  journal={2015 IEEE Winter Conference on Applications of Computer Vision},
  year={2015},
  pages={1046-1053}
}
Truncated signed distance function (TSDF) based volumetric surface reconstructions of static environments can be readily acquired using recent RGB-D camera based mapping systems. If objects in the environment move then a previously obtained TSDF reconstruction is no longer current. Handling this problem requires segmenting moving objects from the reconstruction. To this end, we present a novel solution to the motion segmentation of TSDF volumes. The segmentation problem is cast as CRF-based MAP… 

Figures from this paper

Rigid body motion segmentation with an RGB-D camera

This thesis investigates the problem of rigid body motion segmentation using such a camera, and proposes a novel binary similarity measure and a novel guided sampling scheme termed Rigidity-GS which improves the probability of a data sample being an all-inlier sample.

Static-Map and Dynamic Object Reconstruction in Outdoor Scenes Using 3-D Motion Segmentation

A Sparse subspace clustering-based motion segmentation method that categories the static scene parts and the multiple moving objects using their 3D motion trajectories and outperforms its 2-D-based counterparts, for extensive experiments on KITTI dataset.

Shape-Aware Organ Segmentation by Predicting Signed Distance Maps

By converting the segmentation task into predicting an SDM, this work shows that the proposed model with SDM learning produces smoother segmentation results with smaller Hausdorff distance and average surface distance, thus proving the effectiveness of the method.

Temporally Consistent Motion Segmentation From RGB‐D Video

This work forms segmentation as a motion assignment problem, where a motion is a sequence of rigid transformations through all frames of the input, and captures the quality of each potential assignment by defining an appropriate energy function that accounts for occlusions and a sensor‐specific noise model.

Shape-aware Semi-supervised 3D Semantic Segmentation for Medical Images

This work proposes a novel shapeaware semi-supervised segmentation strategy to leverage abundant unlabeled data and to enforce a geometric shape constraint on the segmentation output and develops a multi-task deep network that jointly predicts semantic segmentation and signed distance map(SDM) of object surfaces.

SimCVD: Simple Contrastive Voxel-Wise Representation Distillation for Semi-Supervised Medical Image Segmentation

SimCVD is presented, a simple contrastive distillation framework that significantly advances state-of-the-art voxel-wise representation learning and hypothesize that dropout can be viewed as a minimal form of data augmentation and makes the network robust to representation collapse.

Multiview point clouds denoising based on interference elimination

This study provides the feasibility of obtaining fine 3-D models with high-noise devices, especially for depth sensors, such as Kinect, with good robustness, compared to truncated signed distance function and moving least squares.

3-D Deep Feature Construction for Mobile Laser Scanning Point Cloud Registration

The experimental results illustrated that the proposed method of using 3-D deep features is more efficient at a corresponding point search than representatives of three existing methods and it also improved registration accuracy.

Deep learning to segment pelvic bones: large-scale CT datasets and baseline models

This paper aims to bridge the data gap by curating a large pelvic CT dataset pooled from multiple sources, and proposes for the first time to learn a deep multi-class network for segmenting lumbar spine, sacrum, left hip, and right hip, from multiple-domain images simultaneously to obtain more effective and robust feature representations.

When Anchoring Fails: Interactive Alignment of Large Virtual Objects in Occasionally Failing AR Systems

Several interaction algorithms are implemented to make manual virtual object alignment easier, based on previously presented methods, such as HoverCam, SHOCam, and a Signed Distance Field, which can reduce the time needed for interactive 3D pose alignment, which improves the user experience.

References

SHOWING 1-10 OF 22 REFERENCES

Dense multibody motion estimation and reconstruction from a handheld camera

A dense solution to all three elements of this problem: depth estimation, motion label assignment and rigid transformation estimation directly from the raw video by optimizing a single cost function using a hill-climbing approach.

Maximal Cliques Based Rigid Body Motion Segmentation with a RGB-D Camera

This paper presents a novel method for motion segmentation without explicitly estimating motions, and shows that identifying true motions from the candidate motions can be cast as a minimum set cover problem or a max k-cover problem for data with outliers.

Real-Time Camera Tracking and 3D Reconstruction Using Signed Distance Functions

This paper presents a novel method for real-time camera tracking and 3D reconstruction of static indoor environments using an RGB-D sensor that is more accurate and robust than the iterated closest point algorithm (ICP) used by KinectFusion, and yields often a comparable accuracy at much higher speed to feature-based bundle adjustment methods such asRGB-D SLAM.

Multi-body Depth-Map Fusion with Non-intersection Constraints

This paper proposes to phrase the multi-body depthmap fusion problem in a convex voxel labeling framework and introduces non-intersection constraints which resolve issues: at any point in time, a point in space can only be occupied by at most one part.

A volumetric method for building complex models from range images

This paper presents a volumetric method for integrating range images that is able to integrate a large number of range images yielding seamless, high-detail models of up to 2.6 million triangles.

KinectFusion: Real-time dense surface mapping and tracking

We present a system for accurate real-time mapping of complex and arbitrary indoor scenes in variable lighting conditions, using only a moving low-cost depth camera and commodity graphics hardware.

KinectFusion: real-time 3D reconstruction and interaction using a moving depth camera

Novel extensions to the core GPU pipeline demonstrate object segmentation and user interaction directly in front of the sensor, without degrading camera tracking or reconstruction, to enable real-time multi-touch interactions anywhere.

Object Segmentation by Long Term Analysis of Point Trajectories

This paper presents a method that uses long term point trajectories based on dense optical flow to define pair-wise distances between these trajectories, which results in temporally consistent segmentations of moving objects in a video shot.

Patch Volumes: Segmentation-Based Consistent Mapping with RGB-D Cameras

Patch volumes is described, a novel multiple-volume representation which enables the creation of globally consistent maps of indoor environments beyond the capabilities of previous high-accuracy volumetric representations.

Fast Approximate Energy Minimization with Label Costs

The main algorithmic contribution is an extension of α-expansion that also optimizes “label costs” with well-characterized optimality bounds, which has a natural interpretation as minimizing description length (MDL) and sheds light on classical algorithms like K-means and expectation-maximization (EM).