【摘要】 为了解决基于内容的图像检索(Content-Based Image Retrieval,CBIR)中存在的"语义鸿沟"问题,本文提出了一种CBIR检索模型,在模型中使用了基于近邻传播的半监督聚类算法和语义传播的算法,通过近邻半监督算法对图像库中的图像进行聚类,根据示例图像的视觉特征相似度在对应的聚类图像中进行相似度检索,在检索的结果中根据用户提供的关键字进行关键字标注检索,最后根据用户的反馈,通过语义传播算法对图像库中的图像进行自动语义标注.实验表明文中的模型是可行的,其检索效果受到反馈次数的影响.更多还原

【Abstract】 To solve the semantic gap in the Content-Based Image Retrieval System,the paper proposes a CBIR model.The Affinity Propagation Clustering Algorithm and semantic propagation algorithm are applied in the model.The Affinity Semi-Supervised Clustering Algorithm clustered the image in the image library.Similarity retrieval in the cluster is carried out through vision feature of the sample image.Keyword retrieval in the retrieval results is carried out through the referring to the keyword.Automatic semantic annotation is realized through semantic propagation.The experimental results indicate the validity of the model and the retrieval effectiveness is affected by the time of feedback.更多还原