计算机科学
点云
点(几何)
人工智能
云计算
数学
几何学
操作系统
作者
Junle Yu,Wenhui Zhou,Zhehao Shen,Yongwei Miao
标识
DOI:10.1016/j.eswa.2025.128591
摘要
• We propose an efficient prior-embedded learning framework for enhancing the pose registration of point clouds, demonstrating significant robustness across both high-overlap and low-overlap scenarios. • The proposed local geometric self-attention module effectively reduces the GPU memory consumption and inference time compared to previous work, while maintaining comparable performance. • The proposed method efficiently integrates various overlap priors to enhance registration performance, such as 2D prior, 3D prior, hybrid 2D/3D prior and random prior. • The proposed method outperforms the state-of-the-art on several public datasets, including 3DMatch/3DLoMatch, ScanNet and KITTI datasets, demonstrating its generality and effectiveness across different datasets. Learning discriminative point-wise features is critical for partially overlapping point cloud registration. In recent years, the integration of a Transformer into point cloud feature representation has demonstrated remarkable success, which typically involves a self-attention module to learn intra-point-cloud features, followed by a cross-attention module for feature exchange between input point clouds. Transformer models mainly benefit from the use of self-attention to capture the global correlations in feature space. However, the global correlations involved in self-attention may not only result in a significant amount of redundant computational overhead but also introduce feature ambiguities, especially in low-overlap scenarios. This is because overlapping regions of point clouds typically do not span a wide range but are rather concentrated around a localized area. Therefore, the correlations with an extensive range of non-overlapping points are ineffective and may degrade the discriminability of features. To address this issue, we present a E fficient P rior- E mbedded A ttention L earning model ( E fficientPEAL). By incorporating overlap prior to the learning process, the point clouds are divided into two parts. One part includes points lying in the putative overlapping region and the other includes points located in the putative non-overlapping region. Then, EfficientPEAL performs localized attention with the putative overlapping points. The proposed attention module significantly reduces the computational complexity of the model while achieving competitive performance. Extensive experiments on 3DMatch/3DLoMatch, ScanNet, and KITTI datasets demonstrate its effectiveness.
科研通智能强力驱动
Strongly Powered by AbleSci AI