【摘要】 视频内容检索是多媒体应用的一个活跃研究方向,现有的内容检索技术大多是基于低层次特征的。这些非语义的低层特征难以理解,与人思维中的高层语义概念相差甚远,严重影响视频内容检索系统的易用性。低层特征和高层语义概念间的语义鸿沟很难逾越。如何跨越语义鸿沟,用语义概念检索视频内容是目前基于内容视频检索最具挑战性的研究方向。本文介绍语义视频检索出现的背景,分析语义鸿沟出现的原因,对现有尝试跨越语义鸿沟的主要方法进行综述;评述了相关技术的优缺点,探讨了各方法将来可能的研究发展方向以及视频语义检索近期、长期可能的技术突破点。更多还原

【Abstract】 Contend-based video retrieval(CBVR)is an active research domain in muhimedia application. Most retrieval techniques on CBVR are low-lever feature based. However,these features are abstract and quite different from the se- mantic concepts in human thought. Video retrieval at semantic level is one of the most challenging research issues in CBVR at present. The gap between low-level features and high semantics is difficult to narrow. To go beyond low-level similarity and access video data content by semantics,we must bridge the gap between the low-level features and high- level semantics. After providing a summary about no-semantic video retrieval in the literature, this paper analyzes the reason leading to semantic gap and presents a review on the current approach to bridging the semantic gap. In addition, the future promising directions are also discussed.更多还原