Development of a New Automatic Relative Radiometric Normalization Methodology for Multispectral and Hyperspectral Images

The concept of a relative radiometric normalization (RRN) for multitemporal multispectral and hyperspectral images is investigated. A new automatic RRN methodology is presented: the Normalized Proximity and Similarity Methodology (NPSM). The NPS methodology introduces two spectral metrics, the absolute normalized difference (AND) and the absolute normalized ratio difference (ANRD), to select Invariant Pixels (IPs). A new metric is also introduced for the evaluation of the proposed methodology: the root mean square error (RMSE) between the values of the initial subject image and the relevant reconstructed subject image. The NPSM is validated using 9 ASTER and 14 Hyperion images. Results demonstrate the high performance of the methodology on both hyperspectral and multispectral data.