English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Zendy Plus

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Zendy Tools

Zendy Open

Point cloud registration aims at estimating the geometric transformationbetween two point cloud scans, in which point-wise correspondence estimation isthe key to its success. In addition to previous methods that seekcorrespondences by hand-crafted or learnt geometric features, recent pointcloud registration methods have tried to apply RGB-D data to achieve moreaccurate correspondence. However, it is not trivial to effectively fuse thegeometric and visual information from these two distinctive modalities,especially for the registration problem. In this work, we propose a newGeometry-Aware Visual Feature Extractor (GAVE) that employs multi-scale locallinear transformation to progressively fuse these two modalities, where thegeometric features from the depth data act as the geometry-dependentconvolution kernels to transform the visual features from the RGB data. Theresultant visual-geometric features are in canonical feature spaces withalleviated visual dissimilarity caused by geometric changes, by which morereliable correspondence can be achieved. The proposed GAVE module can bereadily plugged into recent RGB-D point cloud registration framework. Extensiveexperiments on 3D Match and ScanNet demonstrate that our method outperforms thestate-of-the-art point cloud registration methods even without correspondenceor pose supervision. The code is available at: https://github.com/514DNA/LLT.

Improving RGB-D Point Cloud Registration by Learning Multi-scale Local Linear Transformation