Method for registration of 3D shapes without overlap for known 3D priors
Abstract In 3D registration of point clouds, the goal is to find an optimal transformation that aligns the input shapes, provided that they have some overlap. Existing methods suffer from performance degradation when the overlapping ratio between the neighbouring point clouds is small. So far, there...
Main Authors: | P. Hu, A. Munteanu |
---|---|
Format: | Article |
Language: | English |
Published: |
Wiley
2021-04-01
|
Series: | Electronics Letters |
Subjects: | |
Online Access: | https://doi.org/10.1049/ell2.12133 |
Similar Items
-
SLRP: Improved heatmap generation via selective layer‐wise relevance propagation
by: Yeon‐Jee Jung, et al.
Published: (2021-05-01) -
Constructing an efficient and adaptive learning model for 3D object generation
by: Jiwei Hu, et al.
Published: (2021-06-01) -
Light field editing in the gradient domain
by: Ioana Sevcenco, et al.
Published: (2021-04-01) -
Hyperspectral image classification with deep 3D capsule network and Markov random field
by: Xiong Tan, et al.
Published: (2022-01-01) -
Robust active contours based on local threshold preprocessing fitting energies for fast segmentation of inhomogenous images
by: Baojun Guo, et al.
Published: (2021-07-01)