A5004817418- Computer Graphics and Visualization Techniques
- 3D Shape Modeling and Analysis
- Advanced Data Compression Techniques
- Advanced Vision and Imaging
- Medical Image Segmentation Techniques
- Advanced Numerical Analysis Techniques
- Medical Imaging Techniques and Applications
- Model Reduction and Neural Networks
- Optical Systems and Laser Technology
- Video Coding and Compression Technologies
- Human Pose and Action Recognition
- Advanced MRI Techniques and Applications
- Image and Video Quality Assessment
- Robotics and Sensor-Based Localization
- Optical measurement and interference techniques
- 3D Surveying and Cultural Heritage
Northwestern Polytechnical University
2022-2023
Point cloud compression (PCC) is crucial for efficient and flexible storage as well feasible transmission of point clouds in practice. For geometry compression, one popular approach the octree-based solution. The intra prediction mechanism utilizes spatial correlation static to predict occupancy bit octree node entropy coding, reducing redundancy. In this study, two local geometry-based methods are proposed following statistical theoretical analyses: binary prediction, which outputs state...
Near-lossless compression of point clouds is suitable for the application scenarios with low distortion tolerance and certain requirements on rate. attribute usually adopts a level-of-detail structure, where dependencies between layers make it possible to improve rate-distortion (R-D) performance by using different quantization parameters layers. In this work, theoretical analysis adjacent carried out, based which dependent Distortion-Quantization Rate-Quantization models are established...
Geometry coding in geometry-based point cloud compression (G-PCC) is octree-structured, including a bitwise occupancy mode for general case, and single child node containing occupied node. However, the current usage of limited because strict eligibility determination based on neighboring nodes. Context modeling also missing entropy coordinate index relative to Guided by logistic regression (LR), this paper first proposes an algorithm determine mode. Without resorting nodes, proposed provides...
The Point cloud is a popular representation format of 3D objects and scenes. For efficient transmission storage point clouds in practice, compression becomes an attractive research topic for academia industry. Octree coding one the main features geometry clouds, as employed latest international standard Geometry-based Cloud Compression (G-PCC). This paper aims to improve performance octree G-PCC with reduced complexity. this purpose, we employ neighboring nodes model contexts entropy...
Region adaptive hierarchical transform (RAHT) is employed in geometry-based point cloud compression (G-PCC) to make attribute more efficient. The performance of RAHT closely related the quantization parameter (QP), where applying different QPs depths beneficial for coding efficiency. In this paper, QP cascading (QPC) designed based on rate-distortion analysis and modelling. Firstly, single-layer rate-quantization distortion-quantization models are built by investigating distribution...
Since the geometry constitutes most of bitrate and is used for attribute coding, it crucial geometry-based point cloud compression (G-PCC). However, current research focuses on coding while ignoring reconstruction. In G-PCC, reconstructed points are located at center quantization nodes, which may not match surface clouds. Therefore, we first estimate normal direction. Then, offset direction determined by considering its adjacent points' occupancy attributes. Finally, position adjusted taking...
Linear reduced-order modeling (ROM) simplifies complex simulations by approximating the behavior of a system using simplified kinematic representation. Typically, ROM is trained on input created with specific spatial discretization, and then serves to accelerate same discretization. This discretization-dependence restrictive. Becoming independent discretization would provide flexibility mix match mesh resolutions, connectivity, type (tetrahedral, hexahedral) in training data; novel...
To improve the reconstructed point cloud after adaptive geometry quantization in geometry-based compression, a least squares plane (LSP) projection-based up-sampling method and parameter (QP) decision based on loss function are proposed. First, LSP fitting is carried out to locate interpolated nearest neighbors of current node during decoding, enhancing both subjective objective quality cloud. Second, QP for each mean squared error between original The experimental results show that proposed...