Extraction of information from satellite images, due to better cost & time consumption, all times accessibility and vast area of coverage, is seen as one solution for countries have no overall maps One way of accessing this information is using the classification of satellite images. Nowadays, many experts are working on it and different approach are examined. One of these approaches is the use of fuzzy technique and another is neural network approach to classify satellite images. In this paper we use these techniques. High quickness comparing existing approaches such as classification based on maximum likelihood and high accuracy are from its advantages. Spectral mixture analysis is done by two methods, hyper spectral classification is done with spectral matching technique which is first method and pixel purity analysis with spectral unmixing logarithm is second method. Now spectral similarity can be measured between spectral signatures of image pixel and spectral signature of target pixel using spectral matching technique. Spectral similarity can be measured in different methods. They are divergence, entropy, probability, distance, spectral angle, correlation etc. In my project, spectral matching technique is performed by combining the VISA method and SCM method. The VISA method would detect minute signal variation of spectral data at different wavelength interval. In hyperspectral remote sensing, sensors acquire reflectance values at many different wavelength bands, to cover a complete spectral interval. These measurements are strongly correlated and no new information might be added when increasing the spectral resolution. Moreover, the higher number of spectral bands increases the complexity of a classification task. Therefore, feature extraction is a crucial step. An alternative would be to choose the required sensor bands settings a priori. In this paper, we develop a statistical procedure to provide optimal sensor settings for a classification task at hand. The procedure selects wavelength band settings which optimize the separation between the different spectral classes. The method is applicable as a band reduction technique, but it can as well serve the purpose of data interpretation or sensor design. Results on a vegetation classification task show an improvement in classification performance over feature selection and other band selection techniques.