| Image annotation refers to the labeling of images with a set of predefined keywords. Although an easy task for human beings, image annotation has been proven to be an extremely difficult problem for computers. In this thesis, a novel automatic image annotation system is proposed that integrates two sets of support vector machines (SVMs), namely, the multiple instance learning-based (MIL) and global feature-based SVMs, for annotation. The MIL-based features are obtained by applying MIL on the image blocks, where the modified diversity density (DD) algorithm and a faster searching algorithm are applied to improve the efficiency and accuracy. They are further input to a set of SVMs for finding the optimum hyperplanes to annotate training images. Similarly, global color and texture features, including color histogram and modified edge histogram, are fed into another set of SVMs for categorizing training images. Consequently, two sets of image features are constructed for each test image and are respectively sent to the two sets of SVMs, whose outputs are incorporated according to a novel automatic weight estimation method to obtain the final annotation results. The proposed annotation approach demonstrates a promising performance for an image database of 6000 general-purpose images from COREL, as compared with some current peer systems in the literature. |
