目的 类似于文本摘要,视频摘要是对视频内容的总结。为了合理地评估视频摘要领域的研究进展,正确导向视频摘要的继续研究,本文归纳总结视频摘要技术的主要研究方法和显著性成果,对视频摘要技术进行综述。方法 依据视频摘要的两个主要生成步骤:视频内容分析和摘要生成分别介绍视频摘要的主要研究方法。同时,分析了近5年视频摘要领域的研究状况,对视频摘要发展的新趋势:实时视频摘要和多视角视频摘要进行了阐述。最后,还对视频摘要的评价系统进行了分类总结。结果 对视频摘要进行综述,对摘要中的语义获取难题提出了2种指导性建议。并依据分析结果,展望了视频摘要技术未来的发展方向。结论 视频摘要技术作为视频内容理解的重要组成部分,有较大研究价值。而目前,视频摘要在视频语义表达和摘要评价系统方面并不精确完善,还需进一步的深入研究。

Objective This paper summarized the significant methods and techniques for video abstraction and systematically presented all aspects of this technology to properly evaluate the research progress of video abstraction and to provide a correct orientation for deep research in this field. Method This paper introduced the main research techniques in detail and discussed the typical algorithms based on two main abstraction generation steps, namely, video content analysis and abstraction generation. Specifically, this paper analyzed the progress of video abstraction in the past five years and described the new demand in this field, that is, real-time and multi-view video abstraction. The evaluation benchmark of video abstraction was also analyzed.Result This paper summarized the video abstraction techniques and described the current research situation in this field. Two solutions to address the difficulty in obtaining video semantics were also proposed. Finally, the paper discussed the future development direction of video summary technology.Conclusion As an important part of understanding video content, video abstraction has great research value. To date, video abstraction still has defects that need to be addressed, particularly in the areas of video semantic presentation and abstraction evaluation benchmark.