Locality perserving projections algorithm for hyperspectral image dimensionality reduction

Feature extraction is an necessary preprocessing step for information extraction from hyperspectral remote sensing data. We exploit locality preserving projections (LPP) for dimensionality reduction, leading to a low-dimensional embedding. Given a subset of hyperspectral image, by LPP, we project them into a low-dimensional space to preserve much of the geometric structure. Hyperspectral image classification is then achieved using Support Vector Machine. To evaluate the proposed method, LPP method was examined in terms of spatial information preservation. The preliminary result of this study demonstrated that it is effective.