Large-scale image annotation using prototype-based models

Automatic image annotation is a challenging problem in the field of image retrieval. Dealing with large databases makes the annotation problem more difficult and therefore an effective approach is needed to manage such databases. In this work, an annotation system has been developed which considers images in separate categories and constructs a profiling model for each category. To describe an image, we propose a new feature extraction method based on color and texture information that describes image content using discrete distribution signatures. Image signatures of one category are partitioned using spectral clustering and a prototype is determined for each cluster by solving an optimization problem. The final model of one category will be constructed based on its prototypes in a generative modeling framework. To study the accuracy of the proposed approach, a large amount of pictures from a standard dataset are used for experiment. The results reveal that our system outperforms other methods which follow the same structure as the proposed approach.