Remote sensing of hyperspectral satellite images has shown to have wide applications. One problem that researchers face is the high volume and dimensionality of the data, or the curse of dimensionality. In this paper, we have used the crop stage classification problem to assess the performance of different dimensionality reduction techniques. We have made use of three techniques, Principal Component Analysis (PCA), Independent Component Analysis (ICA) and Vertex Component Analysis (VCA) for the purpose of dimensionality reduction. We then use a Multi-Layer Perceptron Neural Network for the purpose of classification. Results of each scheme are evaluated interms of the classification efficiency and computational complexity afforded. Results obtained indicate that the use of PCA is least computationally intensive, while classification results are better for the data reduced using ICA and VCA.