Resolving the celestial classification using fine k-NN classifier

With the rapid growth in space technology, space exploration is on the high demand. With each such type of mission, data is accumulating in heaps. Be it manned or unmanned mission, its credibility is defined by the quality of research which can be conducted on the data collected in such missions through remote or on the capsule experiments. Thus there is huge demand of soft techniques, which can make the space or celestial data as useful as possible. One of the major issues is dearth of automated technique for image classification of celestial bodies. Though many image classification techniques exist, but none of them is totally attributed to celestial bodies. An artificial neural network based classifier is proposed to classify celestial object from its image. Texture features are extracted from 90 images of size of 225∗225 of different planets. Different classifiers were applied on this training data. Accuracy of different classifiers is compared to find out the best classifier for space data classification. Different validation schemes are applied and the results are compared to figure out the best validation scheme.