Image retrieval based on intrinsic dimension and Shannon entropy
Liang Lei, Tongqing Wang, Jun Peng, Bo Yang · 2011
How to find out a particularly efficient search algorithm in the premise of a considerable accuracy is highlighted in the research of Web content-based image retrieval. This paper focuses on dimensionality reduction and similarity measure of Web image. First, the paper presents the current commercial search engines how to look for Web images. Then, it describes commonly used methods for the non-linear dimension reduction of Web images, follows by proposing intrinsic dimension estimator that is based on HSV features, where the HSV color histogram intersection was used as the function of similarity judgments. And the similarity measure based on Shannon entropy is discussed. Finally, some improvements are made on computing the Shannon mutual information. The results showed that this method has greatly improved the image retrieval in time and precision rates.