Название: Point Cloud Compression: Technologies and StandardizationАвтор: Ge Li, Wei Gao, Wen Gao
Издательство: Springer
Год: 2024
Страниц: 264
Язык: английский
Формат: pdf (true), epub
Размер: 31.1 MB
3D point clouds have broad applications across various industries and have contributed to advancements in fields such as autonomous driving, immersive media, metaverse, and cultural heritage protection. With the fast growth of 3D point cloud data and its applications, the need for efficient compression technologies has become paramount. This book delves into the forefront of point cloud compression, exploring key technologies, standardization efforts, and future prospects.
The fast developments of 3D vision technologies and applications have raised the great interests from both the academia and industry communities. 3D point clouds can have the strong capability of providing immersive visual experience and high-precision 3D modeling of scenes. We have witnessed the significant progress of compression, processing, and analysis technologies for point clouds. Especially, the available limited bandwidth always restricts the storage and transmission of point clouds, and efficient compression and streaming technologies have been explored broadly. Similar to big image and video data, the huge amount of data of point clouds require more efficient compression algorithms to obtain a desirable rate-distortion tradeoff. Standardization organizations (e.g., MPEG, JPEG, AVS, etc.) and corporations have invested many efforts to develop the non-learning and learning methods to better exploit redundancies for compressing the massive point cloud data. The quality improvements from point cloud compression with constrained bandwidth play a critical role in boosting the application utilities of point clouds in the wild.
As a burgeoning format of spatial data, 3D point cloud has triggered a research boom in the field of vision in recent years. It plays an essential role in many modern industrial applications, such as autonomous driving and virtual/augmented reality. To keep up with the ever-growing application requirements and meet the three progressive goals of ease of storage, transmission, and processing, it is of increasing significance to research and develop efficient compression algorithms to deal with the huge amount of point clouds. Point cloud compression technologies have become research hotspots for the past few years. In this chapter, we first introduce the basic concepts of 3D point clouds and the current mainstream point cloud data acquisition methods. After that, we present several public point cloud benchmark datasets and specific application scenarios of point cloud compression techniques. We further elaborate on the development trends and challenges in point cloud technologies.
With the development of multimedia technology, 3D vision has aroused extensive research enthusiasm. The basis of 3D vision is to model the exterior surface of a continuous 3D model. Point cloud is a widely used representation of 3D geometric models. Point cloud compression technologies have become an essential branch of 3D computer vision. Point cloud is a collection of discrete sampling points on underlying surfaces embedded in 3D space. Depending on the point cloud data acquisition equipment, a point cloud comprises geometric information (three-dimensional geometric coordinates x, y, and z) and a series of optional associated attribute information (like normal vector, intensity, color, reflectance, etc.). As a digital representation of a real-world 3D scene, a point cloud is generally used to describe the surface of 3D objects, reflecting the authentic properties of the object surfaces. A point cloud may consist of millions of points for a high-quality representation of a real-world 3D scene. The direct transmission or storage of point clouds has exorbitant requirements on transmission bandwidth and storage capacity, and it is difficult to deploy it in real scenarios. Therefore, it is indispensable to compress point cloud data effectively. Point cloud compression refers to the compression of point cloud data to reduce the occupation of storage space and transmission bandwidth. Point cloud compression technologies employ lossless or lossy compression methods to eliminate redundant information, making it widely applicable in 3D data processing, computer vision, robotics, and other related fields.
This book contains 11 chapters, where the first two chapters give the introduction and background knowledge, and the last chapter discusses the future works of point cloud technologies. Chapters 3–10, which are the main part of this book, focus on the key technologies and standards in point cloud compression. Among these chapters, Chaps. 3–6 present the fundamental knowledge about compression technologies, such as prediction coding, transform coding, quantization techniques, and entropy coding. Chapters 7–10 discuss the MPEG and AVS standards for point cloud compression, such as MPEG V-PCC, MPEG G-PCC, AVS PCC, and the on-going AI-based PCC.
Chapter 1 introduces the basic concepts of point clouds, acquisition equipment, representative datasets, typical application fields, development trends, and challenges.
Chapter 2 mainly focuses on the background knowledge of point cloud compression, which starts with the fundamentals of information theory. Afterward, point cloud compression technologies are generally introduced. Additionally, point cloud quality evaluation methods are also described.
Chapters 3–6 are technical chapters on the main coding tools, including prediction, transform, quantization, and entropy coding for the efficient compression of point clouds.
Chapter 7 describes the MPEG standard for geometry-based point cloud compression (MPEG G-PCC), including the coding scheme, geometry coding, attribute coding, quantization, entropy coding techniques, and emerging standard activities.
Chapter 8 presents the AVS standard for point cloud compression (AVS PCC), which begins with a brief introduction of the AVS point cloud compression standard timeline and synopsis. Afterward, the geometry octree, geometry predictive tree, attribute prediction, attribute transform, and attribute entropy coding are discussed in detail.
Chapter 9 introduces the MPEG standard for video-based point cloud compression (MPEG V-PCC), including the encoding and decoding frameworks of V-PCC, point cloud projection approach, and projected video compression methods.
Chapter 10 focuses on the deep learning-based point cloud compression methods, including the basic principles, the related neural network design and optimization methods, and the MPEG AI-3DGC standard. Note that the China AVS working group has established a group working on the standardization efforts for AI-based PCC in 2023.
Chapter 11 summarizes future works on point cloud technologies from different technical perspectives, including the main coding tools, MPEG G-PCC standard, AVS point cloud compression standard, MPEG V-PCC standard, and deep Learning-based point cloud compression technologies.
This book analyzes and presents the advanced point cloud compression technologies, which have shown significant progresses in recent years. As a textbook on 3D point cloud compression technologies and standardization, this book comprises the above selected chapters. Through the progressive presentation, readers can comprehensively understand and master the basic knowledge, the main techniques, and the development trends of point cloud compression. To the best of our knowledge, this is the first comprehensive compilation of all aspects of point cloud compression technologies, as well as the standardization efforts. We hope you enjoy this book and join the growing community of point cloud compression enthusiasts.
Contents:
Скачать Point Cloud Compression