摘要
随着对电视节目素材和新闻、广告等节目监播检索需求的不断增加,基于文本方式的检索已不能满足要求。基于颜色特征的电视图像检索能够将电视图像检索从原始的文本检索提升到基于内容的检索,通过其原理和算法分析,并用Matlab对广告、建筑、人物等4类图像进行仿真检索实验,结果表明,基于颜色特征的电视图像检索是可行的,在实验的4类图像中都能准确地检索出目标图像。
With the increasing of TV program material retrieval and news, advertisement retrieval requirements, the text retrieval can＇t meet the need. The TV image retrieval based on color feature can raise the level of TV image retrieval from primitive text retrieval to contented-based image retrieval. By the principle and algorithm analysis, and Matlab retrieval experiment in four different kinds of image, such as advertisement, building and portrait etc., can be resulted that the four sample images all can be accurately found out from the experiment images. The TV image retrieval based on color feature is feasible.
出处
《电视技术》
北大核心
2012年第8期68-71,共4页
Video Engineering
关键词
颜色直方图
相似性匹配
电视图像检索
性能评价
color histogram
similarity match
TV image retrieval
performance evaluation
