
320 Supervised Classification Part 2
FIGURE 7.9
Anomaly detection: Top, RGB composite of bands 3, 2, and 1 of a spatial
subset of the 9- band pan-sharpened ASTER image over J¨ulich. Bottom left,
RX anomaly image calculated from all 9 bands using the script in Lis ting 7.2.
Bottom right, kernel RX a nomaly image calculated from all 9 bands using the
ENVI/IDL extension KRX_RUN.PRO. (See color insert.)
7.6 Exercises
1. Derive the confidence limits on the misclass ific ation rate θ given by
Equation (7.14).
2. In bootstrap aggregation or bagging (Breiman, 1996; Polikar, 2006) an
ensemble of classifiers is derived from a sing le classifier by training it re-