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/

Global: Zendy Plus annual

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

Global: Zendy Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Point cloud registration is a fundamental task in many applications such aslocalization, mapping, tracking, and reconstruction. Successful registrationrelies on extracting robust and discriminative geometric features. Thoughexisting learning based methods require high computing capacity for processinga large number of raw points at the same time, computational capacitylimitation is not an issue thanks to the powerful parallel computing processusing GPU. In this paper, we introduce a framework that efficiently andeconomically extracts dense features using a graph attention network for pointcloud matching and registration (DFGAT). The detector of the DFGAT isresponsible for finding highly reliable key points in large raw data sets. Thedescriptor of the DFGAT takes these keypoints combined with their neighbors toextract invariant density features in preparation for the matching. The graphattention network (GAT) uses the attention mechanism that enriches therelationships between point clouds. Finally, we consider this as an optimaltransport problem and use the Sinkhorn algorithm to find positive and negativematches. We perform thorough tests on the KITTI dataset and evaluate theeffectiveness of this approach. The results show that this method with theefficiently compact keypoint selection and description can achieve the bestperformance matching metrics and reach the highest success ratio of 99.88%registration in comparison with other state of the art approaches.

Learning Dense Features for Point Cloud Registration Using a Graph Attention Network