Enhancing content-based image retrieval using machine learning techniques

摘要: In this paper, we propose a term selection model to help select terms in the documents that describe images improve content-based image retrieval performance. First, introduce general feature model. Second, present painless way for training document collections, followed by selecting and ranking using Kullback-Leibler Divergence. After that, learn classification method, test it on result. Finally, setup series of experiments confirm is promising. Furthermore, suggest optimal values number maxK tuning combination parameter α experiments.