Abstract:Three-dimensional (3D) point clouds have a wide range of applications in the field of 3D vision. The quality of the acquired point cloud data considerably impacts the subsequent work of point cloud processing. Due to the sparsity and irregularity of point cloud data, processing point cloud data has always been challenging. However, existing deep learning-based point cloud dense reconstruction methods suffer from excessive smoothing of reconstruction results and too many outliers. The reason for this is that it is not possible to extract features for local and global features at different scales and provide different levels of attention to different regions in order to obtain long-distance dependence for dense reconstruction. In this paper, we use a parallel multi-scale feature extraction module based on graph convolution and an upsampling method with an added multi-head attention mechanism to process sparse and irregular point cloud data to obtain extended point clouds. Specifically, a point cloud training patch with 256 points is inputted. The PMS module uses three residual connections in the multi-scale feature extraction stage. Each PMS module consists of three parallel DenseGCN modules with different size convolution kernels and different averaging pooling sizes. The local and global feature information of the augmented receptive field is extracted efficiently. The scale information is obtained by averaging the different pooled augmented receptive fields. The scale information was obtained using the different average pooled augmented receptive fields. The upsampling stage uses an upsampling rate of r=4, The self-attentive features with a different focus on different point cloud data regions obtained by fusing different weights make the feature representation more diverse. This operation avoids the bias of one attention, and each focuses on extracting valuable fine-grained feature information. Finally, the coordinate reconstruction module obtains 1024 dense point cloud data. Experiments show that the proposed method demonstrates good evaluation metrics and performance and is able to obtain better visual quality. The problems of over-smoothing and excessive outliers are effectively mitigated, and the obtained sparse point cloud is more dense.

A Two-Stage Transformer-Based Point Cloud Upsampling

PU-Transformer: Point Cloud Upsampling Transformer

Point Cloud Upsampling via Cascaded Refinement Network

Point Cloud Upsampling Algorithm: A Systematic Review

Point cloud upsampling via a coarse-to-fine network with transformer-encoder

Point Cloud Upsampling Via Disentangled Refinement

Transformer-based Point Cloud Generation Network

PU-CTG: A Point Cloud Upsampling Network Using Transformer Fusion and GRU Correction

HFCI-PU: High-Frequency Compensation Interpolation Network for Point Cloud Upsampling

SSPU-Net: A Structure Sensitive Point Cloud Upsampling Network with Multi-Scale Spatial Refinement

Upu-Snet: Siamese Network for Unsupervised Point Cloud Upsampling Based on Spatial-Aware Transformers

Data-driven Upsampling of Point Clouds

Refine-PU: A Graph Convolutional Point Cloud Upsampling Network using Spatial Refinement

Grad-PU: Arbitrary-Scale Point Cloud Upsampling via Gradient Descent with Learned Distance Functions

Transformer and Upsampling-Based Point Cloud Compression

PU-Net: Point Cloud Upsampling Network

PU-EdgeFormer: Edge Transformer for Dense Prediction in Point Cloud Upsampling

Soft Masked Transformer for Point Cloud Processing with Skip Attention-Based Upsampling

APUNet: Attention-guided Upsampling Network for Sparse and Non-Uniform Point Cloud

Neural Points: Point Cloud Representation with Neural Fields for Arbitrary Upsampling

Sparse 3D Point Cloud Parallel Multi-Scale Feature Extraction and Dense Reconstruction with Multi-Headed Attentional Upsampling