Abstract: Latent semantic analysis is based on video analysis. It constructs the video feature matrix via a kind of mapping and realizes video based content retrieval. The paper discusses the main idea of latent semantic analysis and researches the method color texture extraction. The result proves latent semantic analysis performs preferable in content based video retrieval.

Key words: content based video retrieval, latent semantic analysis, singular value decomposition

摘要： 潜在语义分析技术是建立在视频分析基础之上的，它通过某种映射关系来建立视频特征矩阵，实现了基于内容的视频检索。阐述了潜在语义分析技术，进行了视频颜色和纹理特征提取研究，实验结果表明，潜在语义分析对于视频内容检索效果较好。

关键词: 基于内容的视频检索, 潜在语义分析, 奇异值分解

CLC Number: 

• TP37