基于虚拟关联反馈与聚类的图像检索技术研究

Research on pseudo-relevance feedback and clustering-based image retrieval

基于虚拟相关反馈(PRF)技术,提出了一种新的自动关联反馈检索方法——外部自动相关反馈(OARF)。该方法基于图像内容特征距离,应用K-均值聚类,自动扩展查询图像特征,从而提高检索性能。试验结果表明,OARF能够降低用户负担,显著提高原始检索算法的性能,缩小"语义鸿沟"。

A novel auto relevance feedback image retrieval method called the outer auto relevance feedback（OARF） is proposed based on the pseudo-relevance feedback technology.This method selects positive images according to the feature distance of images from the initial retrieval results using the K-means clustering approach.Retrieve images to each of the positive images which cover more objective features.Experimental results demonstrate that the OARF can release users＇ burdens in retrieval and improve the retrieval performance of the original algorithms greatly and then narrows the ＂semantic gap＂.