点击次数：

DOI码：10.1016/j.neucom.2025.132575

发表刊物：Neurocomputing

关键字：Audio-visual fusion 音频视频融合 Deep learning 深度学习 Multimodal 多模态 Survey 综述

摘要：With the development of society, audio and video have become predominant forms of media in our daily lives. Current audio-visual fusion (AVF) survey papers are classified according to fusion stages or application scenarios. Although they introduce the development and analyze the performance of different AVF-based methods, the survey papers overlook the impact of fundamental AVF techniques. In this paper, we review the development of AVF-based methods in terms of their fusion techniques and application scenarios. Meanwhile, we provide a comprehensive survey on three major AVF-based applications. Furthermore, we present the results of representative algorithms and analyze their performance to build a deeper understanding of these methods. Finally, we summarize the ongoing challenges and issues in the AVF domain and offer insights into future research problems and directions. We aim to provide a fine-grained classification of AVF and serve as a comprehensive reference for researchers in the AVF domain.

论文类型：期刊论文

文献类型：J

卷号：671,132575

是否译文：否

发表时间：2026-03-31

收录刊物：SCI