A Framework for Semantic Image Annotation Using LEGION Algorithm

A new method for the annotation of multispectral satellite images based on image segmentation is proposed in this paper. This method performs the multispectral image annotation by incorporating a modified locally excitatory globally inhibitory oscillatory network (LEGION) algorithm and cascaded support vector machine (SVM) classifier. Initially, images in the training set are represented with semantic concepts. The testing image is segmented into various image regions based on the color information. Segmented image regions are classified using cascaded SVM classifier based on the probabilities of semantic classes. Experiments are conducted on multispectral images of Coimbatore, Tamil Nadu, India and the result validates the effectiveness of the proposed image annotation algorithm.