This preview has intentionally blurred sections. Sign up to view the full version.View Full Document
Unformatted text preview: 2370 IEEE TRANSACTIONS ON IMAGE PROCESSING, VOL. 18, NO. 10, OCTOBER 2009 Content Based Image Retrieval Using Unclean Positive Examples Jun Zhang and Lei Ye , Senior Member, IEEE Abstract— Conventional content-based image retrieval (CBIR) schemes employing relevance feedback may suffer from some problems in the prac- tical applications. First, most ordinary users would like to complete their search in a single interaction especially on the web. Second, it is time con- suming and difficult to label a lot of negative examples with sufficient va- riety. Third, ordinary users may introduce some noisy examples into the query. This correspondence explores solutions to a new issue that image retrieval using unclean positive examples. In the proposed scheme, mul- tiple feature distances are combined to obtain image similarity using clas- sification technology. To handle the noisy positive examples, a new two- step strategy is proposed by incorporating the methods of data cleaning and noise tolerant classifier. The extensive experiments carried out on two different real image collections validate the effectiveness of the proposed scheme. Index Terms— Classifier combination, content-based image retrieval (CBIR), feature aggregation, noise tolerant, support vector machine (SVM). I. INTRODUCTION Content-based image retrieval (CBIR) is a technique to search for images relevant to the user’s query from an image collection . In the last decade, the conventional CBIR schemes employing relevance feedback have achieved certain success . The idea of relevance feed- back is to involve the user in the retrieval process so as to improve the final retrieval results. Normally, the user labels some returned images as relevant or irrelevant and the system adjusts the retrieval parame- ters based on the user’s feedback. Relevance feedback can go through one or more iterations until the user is satisfied with the results. How- ever, the conventional CBIR schemes employing relevance feedback may suffer from some problems in practical applications. First, if not impossible, ordinary users have little patience to persist in the feedback iterations, and most would like to complete their search in a single inter- action . Second, labeling some positive (relevant) examples is easy while labeling sufficient negative (irrelevant) examples is time con- suming and difficult . Third, some noisy examples may present since ordinary users normally have no expertise in constructing a high quality query. To the best of our knowledge, most existing retrieval schemes fail to address the problem of noisy examples. In this correspondence, we explore solutions to a new issue that image retrieval using unclean positive examples. The user supplies several unclean positive examples as a query and the CBIR system will return the relevant images from an image collection in a single interaction. Under this circumstance, some noisy positive examples may present in the query which are irrelevant...
View Full Document
This note was uploaded on 11/03/2009 for the course COMPUTERS CS537 taught by Professor Salman during the Spring '09 term at Texas A&M University–Commerce.
- Spring '09