Image Indexing and Retrieval using Cross-Entropy Measures

| In content-based image indexing and retrieval, images are indexed based on the image attributes like, color, texture, shape. These attributes are used in enabling image retrieval based on query images. A user provides a sample image and the images in the database that are similar to the sample image are retrieved. Image content is represented in the form of feature vectors or histograms. In this paper, color content of the image represented in the form of probability density functions is used as image signatures for image indexing and retrieval. Similarity measure based on cross-entropy measure, speciically Kullback-Leibler measure, de-ned on probability density functions is used for comparing the signature of the images. Some results of image indexing and retrieval are presented.