While applying PCA or LDA on hyperspectral image what is the input vectors of each class.Are we taking the input samples(as vectors) of a class from different wavelength bands or we take an image of a particular band and take the reflectance values of a particular class as input vector. Please clarify my doubt.