Automatic Classification of Aerial Imagery

The aerial imagery obtained from reconnaissance platform is voluminous and the defenceforces rely on image information to perform intelligent tasks. The application of a welldesigned automatic image classifier would enhance the end results of different high levelapplications thereby abridging the effort of a human analyst. Automatic image classifierscould be designed using a training data set for supervised learning or using an unsupervisedlearning. In this paper, a method, which combines both unsupervised and supervised methodsof learning is proposed