摘要
使用基于文本的互联网图像检索技术是互联网图像检索最实用的方式,也对其他方式的互联网图像检索有重要辅助作用,但如何利用周边文本来对图像进行准确描述一直是一个难题。利用TFIDF为基础提出了一个基于句法和文本重要性分类的图像关键词权重计算方法,并尝试通过图像的相似性因素作为反馈进一步优化搜索结果,为用户返回最贴切的搜索结果。
Text-based image retrieval is the most practical way for Internet's image retrieval as well as very important to other image retrieval methods as auxiliary function, but how to use the surrounding text to describe the image accurately has always been a problem.Using TFIDF method as base, this paper puts forward an image's keywords weight method based on syntactic and the text importance classification and tries to further optimize the search results through the similary factors of images as feedback, so that it can return the most relevant search results to the users.
出处
《计算机工程与应用》
CSCD
北大核心
2011年第32期186-190,共5页
Computer Engineering and Applications
基金
湖南省自然科学基金(No.07555084)
广东省科技计划项目(No.2009B080701031)
关键词
图像
文本提取
相似图像匹配
image
text extraction
similar image matching