Research And Implementation Of A IMage Retrieval System Based On Feedback And Multi-Feature

With the popularity of the Internet and mobile internet so much digital imageinformation appears. It becomes a urgent problem for users to choose what they wantfrom it accurately and rapidly. The traditional image retrieval that based on key wordscan not meet their need. So, a new image retrieval based on content come into being.Compared to the former, the new one has two advantages. The one is that it is moreefficient and easy. The other is that it is more accurate. As we all know different peoplehave different understanding and point about the picture. The keywords-based imageretrieval makes it subject and incomplete. The content-based image retrieval presentsthe image by extracting the underlying features. Those underlying features makes itmore objective and accurate in decrypting the picture. Some systems are based onfeature-based image retrieval, some can not give feedback and some can not beexpanded feasibly in existing image retrieval system. The paper presents a new imageretrieval that based on multi-feature feedback.The paper focuses on several parts.1It makes comment on the background andsignificance of CBIR, discusses the basic principles of the image retrievaltechnology,analyzes the drawback of traditional image retrieval system framework andputs up a retrieval system based on web.2It introduces the color features, texturefeatures,shape features and related algorithms in feature-based image retrieval, thesimilarity of image, multi-feature technology and related feedback technology.3Itmakes it more stable and scalable by designing the system based on demand.4It designsthe analysis and extraction of image features and elaborate on algorithm. And finish theextraction and analysis of basic image features.5At last, it designs and finishes theimage retrieval based on multi features. The system is based on web. The system canretrieve by color,shape and texture. And it do retrieval by various traits and changeweight ration based on customers’ feedback,making it more easy and accurate forcustomer to retrieve.