Integrating Relevance Feedback Techniques for Image Retrieval Using Reinforcement Learning
IEEE Transactions on Pattern Analysis and Machine Intelligence
| Hi-index | 0.00 |
This paper proposes a content-based image retrievalsystem which can learn visual concepts and refine themincrementally with increased retrieval experiences captured over time. The approach consists of using fuzzyclustering for learning concepts in conjunction with statistical learning for computing "relevance" weights offeatures used to represent images in the database. Asthe clusters become relatively stable and correspond tohuman concept distribution, the system can yield fastretrievals with higher precision. The paper presentsdiscussion on problems such as system mistakenly identifying a concept, large number of trials to achieveclustering, etc. The experiments on synthetic data andreal image database demonstrate the efficacy of thisapproach.