Semi-automatic image annotation using sparse coding

Automatically assigning keywords to images is of great interest as it allows one to index, retrieve, and understand large collections of image data. It has become a new research focus and many techniques have been proposed to solve this problem. In this paper, a novel semi-auto image annotation technique is proposed. The new developed method uses a label transfer mechanism to automatically recommend promising tags to each image by assigning each image a category label first. Since image representation is one of the key problems in image annotation, we utilize a sparse coding based spatial pyramid matching as an effective way to model and interpret image features. Experimental results demoustrate that the proposed method outperforms the current state-of-the-art methods on two benchmark image datasets.