The human eye can perceive information from the visible light in terms of bands of three colors (red, green, blue), so generally images store in the digital are made up of three dimensions i.e., R, G and B. But hyper spectral imaging perceives information from across the electromagnetic spectrum; the process of spectral imaging further splits the spectrum into more bands. This process of changing images into bands can be even in the invisible spectrum. Hence the hyper spectral images can be considered as n-dimensional matrices and each pixel can be regarded as n-dimens ional vector. These images contain various areas with similar characteristics like crop fields, forest area and deserts. To classify such regions one has look for certain features among the captured images. Some similarity measures should be undertaken to make clusters of areas having similar characteristics from the images. Finding the relative similarities in terms of numerical score can be carried out with the help of some standard algorithm. So, feature classification on basis of relative similarities pixel is robust method. In this paper proposing classification of hyper spectral images using Multilayer Perceptron Artificial Neural Network (MLPANN) and Functional Link Artificial Neural Network (FLANN) and their performance is compare in term of accuracy rate. © 2017 IEEE.