一种基于结构学习的图像标注方法

Image annotation plays an important role in modern keyword-based image retrieval systems. In this paper, we formulate the image annotation problem as that of selecting the optimal keyword subset for an image, and employ a structural learning framework to learn a scoring function for evaluating the quality of different candidate subsets. The quality of a keyword subset is assessed based on its relations with visually similar neighbors of that image. Experiments on benchmark data set demonstrate the effectiveness of our approach for image annotation.