TY - JOUR

T1 - Point Cloud Completion Via Skeleton-Detail Transformer

AU - Zhang, Wenxiao

AU - Zhou, Huajian

AU - Dong, Zhen

AU - Liu, Jun

AU - Yan, Qingan

AU - Xiao, Chunxia

PY - 2023/10/1

Y1 - 2023/10/1

N2 - Point cloud shape completion plays a central role in diverse 3D vision and robotics applications. Early methods used to generate global shapes without local detail refinement. Current methods tend to leverage local features to preserve the observed geometric details. However, they usually adopt the convolutional architecture over the incomplete point cloud to extract local features to restore the diverse information of both latent shape skeleton and geometric details, where long-distance correlation among the skeleton and details is ignored. In this work, we present a coarse-to-fine completion framework, which makes full use of both neighboring and long-distance region cues for point cloud completion. Our network leverages a Skeleton-Detail Transformer, which contains cross-attention and self-attention layers, to fully explore the correlation from local patterns to global shape and utilize it to enhance the overall skeleton. Also, we propose a selective attention mechanism to save memory usage in the attention process without significantly affecting performance. We conduct extensive experiments on the ShapeNet dataset and real-scanned datasets. Qualitative and quantitative evaluations demonstrate that our proposed network outperforms current state-of-the-art methods.

AB - Point cloud shape completion plays a central role in diverse 3D vision and robotics applications. Early methods used to generate global shapes without local detail refinement. Current methods tend to leverage local features to preserve the observed geometric details. However, they usually adopt the convolutional architecture over the incomplete point cloud to extract local features to restore the diverse information of both latent shape skeleton and geometric details, where long-distance correlation among the skeleton and details is ignored. In this work, we present a coarse-to-fine completion framework, which makes full use of both neighboring and long-distance region cues for point cloud completion. Our network leverages a Skeleton-Detail Transformer, which contains cross-attention and self-attention layers, to fully explore the correlation from local patterns to global shape and utilize it to enhance the overall skeleton. Also, we propose a selective attention mechanism to save memory usage in the attention process without significantly affecting performance. We conduct extensive experiments on the ShapeNet dataset and real-scanned datasets. Qualitative and quantitative evaluations demonstrate that our proposed network outperforms current state-of-the-art methods.

U2 - 10.1109/TVCG.2022.3185247

DO - 10.1109/TVCG.2022.3185247

M3 - Journal article

VL - 29

SP - 4229

EP - 4242

JO - IEEE Transactions on Visualization and Computer Graphics

JF - IEEE Transactions on Visualization and Computer Graphics

SN - 1077-2626

IS - 10

ER -