This paper presents a hybrid classification approach combining the use of supervised classification and unsupervised clustering algorithms. The main idea is to reduce the training set by selecting the most appropriated samples for classification by means of K-nearest neighbor (KNN) algorithm. Indeed, for each class the resulting center clusters from Kmeans are chosen as those samples. Experiments are carried out using two well-known databases: Indian Pines, acquired by AVIRIS sensor; and Pavia University, acquired by ROSIS sensor. Results show the efficiency of our proposed approach which significantly reduces the time required in the classification step while the effectiveness/accuracy is kept close to the ones of the original KNN.