场景三维空间感知建模和基于先验的内容重现一直是信息技术围绕人类交互体验的重要努力方向，也是元宇宙和增强现实等数字和现实世界纽带构建技术的核心基础。当前，随着深度学习驱动的信息技术快速发展，特别是神经辐射场兴起，作为建模和内容重现的核心—场景表征技术得到了跨越式发展。本文首先总结了不同表征技术的应用背景和近期基于深度学习所取得的进展；其次，对神经辐射场近期的重要成果进行了梳理和阐述；然后，就神经辐射场的场景构建和交叉领域研究的分析，揭示了显性空间和语义挖掘对神经辐射场的重要价值；最后，结合近期显性空间和语义挖掘中的重要发展方向，单视图3D场景感知，面临的挑战和神经辐射场相关研究展现的裨益，揭示了基于神经辐射场对场景三维空间感知建模和基于先验的内容重现发展带来的机遇。

Three-dimensional scene modeling and prior-based content reproduction have long been cornerstones of human-centered information technology endeavors. Furthermore, they serve as the technological foundation for new technology ideas, such as meta-universe and augmented reality, which integrate the digital and physical worlds. Recently, due to the rapid development of deep learning-driven technology, scene representation technologies have attained rapid achievements, especially with the growth of the neural radiation field (NeRF). This paper begins by reviewing the different representation technologies and is followed by the hot spot, NeRF. Secondly, the recent important results of NeRF are sorted out and described. Thirdly, based on the analysis of scenario building and cross-field research, the importance of explicit space and semantic mining for NeRF is disclosed. Finally, given the virtues brought by the latest achievements from NeRF to the challenges of single-view 3D scene perception, the opportunities for the development of scene modeling and content reproduction technologies with NeRF are revealed.