دانلود کتاب روش‌ها، استانداردها و کاربردهای کدگذاری تصویر و ویدئو مبتنی بر هوش مصنوعی

From high-resolution imaging sensors to immersive media systems, the rapid evolution of multimedia technologies has intensified the demands for efficient visual data compression for the benefits of human and machine perception. While traditional hybrid image and video coding frameworks have served as the cornerstone for the past over three decades, their diminishing marginal returns in compression efficiency now confront fundamental limitations in the achievable coding gains and the computational complexity. This critical juncture motivates our examination of the deep learning’s transformative potential in redefining compression paradigms. This book systematically investigates the image and video data coding approaches driven by the fast-growing artificial intelligence (AI) technologies, structured through 12 interconnected chapters. The organizational logic progresses from foundational principles to implementation challenges, adhering to three core objectives: (1) To establish theoretical bridges between information theory and neural compression methods. (2) To analyze architectural innovations surpassing the conventional codec performance boundaries. (3) To address practical deployment considerations in standardization and application systems. This book illustrate the theories, methods, and applications of AI-based image and video coding. It provides a comprehensive knowledge of image and video coding powered with AI, including fundamental principles, human visual perception modeling, deep learning-based coding methods, advanced coding problems, technical standards, implementations and open-source projects, and future works. The detailed chapters are organized as follows. Chapters 1 and 2 begin with an introduction that lays the foundation for the book. In particular, Chap. 1 outlines the core concepts of image and video coding, and Chap. 2 performs a systematic analysis of the deep learning theoretical framework highly relevant to this research field. Chapter 3 introduces the fundamentals of quality assessment and perceptual modeling, which are essential in evaluating the image and video coding performances, and guiding perceptual optimization in coding processes. Chapters 4 and 5 systematically investigate the utilization of deep neural networks in image and video coding, respectively, presenting the methodologies of using devised techniques to improve coding efficiency. Chapters 6–8 discuss several advanced coding problems, focusing on 3D and multimodal coding, human and machine perception oriented coding, and compression artifacts removal, respectively. Chapter 9 presents the recent standardization efforts on image and video coding from the working groups of different organizations, e.g., IEEE, MPEG, and JPEG, as well as the comparative analysis of emerging standards. Chapter 10 provides the basic knowledge for implementing AI-based image and video coding techniques, including insights into hardware and software considerations. Chapter 11 gives a review of valuable open source projects on image and video coding. Chapter 12 looks to the future, elaborating on potential directions for research and development in AI-based image and video coding technologies. These chapters constitute the whole book, including the fundamental principles and advanced methods for image and video coding with the aid of deep learning. We aim to provide readers with the indispensable knowledge and tools they need to contribute to this dynamic and evolving field, and also aim to accelerate the transition from the theoretical neural compression models to industrially deployable solutions.