Relevance Feedback-based Natural Image Retrieval Using DCD
Liu Deng-sheng · 2007
Content-based image retrieval is a key technique in multimedia.Normally,traditional CBIR system describes and compares images by using low-level physical features,and the international standard MPEG-7 is the novel method whichcan extract and express content features effectively.In this paper,MPEG-7 Dominant Color Descriptor(DCD) is used to extract and express image features.It puts forward a new retrieval method using DCD,and proposes techniques to improve the effectiveness of Relevance Feedback(RF) based image retrieval using DCD. The retrieving is achieved.The data set composed of 1350 digital images is experimented with,and the experiment results have been analyzed,some of which are satisfied,and the precision based on content image retrieving has been improved.