Leveraging Radar Features to Improve Point Clouds Segmentation with Neural Networks

2020 ◽  
pp. 119-131
Author(s):  
Alessandro Cennamo ◽  
Florian Kaestner ◽  
Anton Kummert
Keyword(s):  
Neural Networks ◽  
Point Clouds

Discriminative and generative models for anatomical shape analysis on point clouds with deep neural networks

2021 ◽  
Vol 67 ◽  
pp. 101852
Author(s):  
Benjamín Gutiérrez-Becker ◽  
Ignacio Sarasua ◽  
Christian Wachinger
Keyword(s):  
Neural Networks ◽  
Shape Analysis ◽  
Deep Neural Networks ◽  
Point Clouds ◽  
Generative Models

scholarly journals StickyPillars: Robust and Efficient Feature Matching on Point Clouds using Graph Neural Networks

2021 ◽  
Author(s):  
Kai Fischer ◽  
Martin Simon ◽  
Florian Olsner ◽  
Stefan Milz ◽  
Horst-Michael Gros ◽  
...  
Keyword(s):  
Neural Networks ◽  
Feature Matching ◽  
Point Clouds ◽  
Graph Neural Networks

scholarly journals Fast Quaternion Product Units for Learning Disentangled Representations in SO(3)

2022 ◽  
Author(s):  
Shaofei Qin ◽  
Xuan Zhang ◽  
Hongteng Xu ◽  
Yi Xu
Keyword(s):  
Neural Networks ◽  
Real World ◽  
Message Passing ◽  
Point Clouds ◽  
Rotation Group ◽  
Structured Data ◽  
Data Indexing ◽  
Basic Module ◽  
Real Neuron ◽  
Rotation Groups

Real-world 3D structured data like point clouds and skeletons often can be represented as data in a 3D rotation group (denoted as $\mathbb{SO}(3)$). However, most existing neural networks are tailored for the data in the Euclidean space, which makes the 3D rotation data not closed under their algebraic operations and leads to sub-optimal performance in 3D-related learning tasks. To resolve the issues caused by the above mismatching between data and model, we propose a novel non-real neuron model called \textit{quaternion product unit} (QPU) to represent data on 3D rotation groups. The proposed QPU leverages quaternion algebra and the law of the 3D rotation group, representing 3D rotation data as quaternions and merging them via a weighted chain of Hamilton products. We demonstrate that the QPU mathematically maintains the $\mathbb{SO}(3)$ structure of the 3D rotation data during the inference process and disentangles the 3D representations into ``rotation-invariant'' features and ``rotation-equivariant'' features, respectively. Moreover, we design a fast QPU to accelerate the computation of QPU. The fast QPU applies a tree-structured data indexing process, and accordingly, leverages the power of parallel computing, which reduces the computational complexity of QPU in a single thread from $\mathcal{O}(N)$ to $\mathcal {O}(\log N)$. Taking the fast QPU as a basic module, we develop a series of quaternion neural networks (QNNs), including quaternion multi-layer perceptron (QMLP), quaternion message passing (QMP), and so on. In addition, we make the QNNs compatible with conventional real-valued neural networks and applicable for both skeletons and point clouds. Experiments on synthetic and real-world 3D tasks show that the QNNs based on our fast QPUs are superior to state-of-the-art real-valued models, especially in the scenarios requiring the robustness to random rotations.<br>

scholarly journals Automatic Regularization of TomoSAR Point Clouds for Buildings Using Neural Networks

Sensors ◽  
2019 ◽  
Vol 19 (17) ◽  
pp. 3748 ◽  
Author(s):  
Siyan Zhou ◽  
Yanlei Li ◽  
Fubo Zhang ◽  
Longyong Chen ◽  
Xiangxi Bu
Keyword(s):  
Neural Networks ◽  
Urban Areas ◽  
Three Dimensional ◽  
Point Clouds ◽  
Building Structures ◽  
Processing Efficiency ◽  
Remote Sensing Technique ◽  
Cloud Processing ◽  
Urban Buildings ◽  
Data Segmentation

Tomographic SAR (TomoSAR) is a remote sensing technique that extends the conventional two-dimensional (2-D) synthetic aperture radar (SAR) imaging principle to three-dimensional (3-D) imaging. It produces 3-D point clouds with unavoidable noise that seriously deteriorates the quality of 3-D imaging and the reconstruction of buildings over urban areas. However, existing methods for TomoSAR point cloud processing notably rely on data segmentation, which influences the processing efficiency and denoising performance to a large extent. Inspired by regression analysis, in this paper, we propose an automatic method using neural networks to regularize the 3-D building structures from TomoSAR point clouds. By changing the point heights, the surface points of a building are refined. The method has commendable performance on smoothening the building surface, and keeps a precise preservation of the building structure. Due to the regression mechanism, the method works in a high automation level, which avoids data segmentation and complex parameter adjustment. The experimental results demonstrate the effectiveness of our method to denoise and regularize TomoSAR point clouds for urban buildings.

Sign in / Sign up

Export Citation Format

Share Document