Study on the use of complexity measures for estimation of correctclassification percentage in hyperspectral imagery

摘要

This study presents image complexity measures applied to hyperspectral images and their relation to the percentage ofcorrect classification (PCC). Specifically, it studies the relationship between these metrics and the PCC for MaximumLikelihood and Angle Detection classifiers. First, many complexity measures were studied to determine if there was arelation between the measure and the PCC. Results showed a correlation of above 0.7 between complexity measuresbased on entropy and uncertainty and the PCC of the classifiers mentioned above. Once the relation was established,PCC estimators based on the metrics using simple and multiple regression models were designed. This design wasperformed using data from both synthetic and real images. The real images were from two hyperspectral sensors, thespace based AISA and a portable SOC 700 hyperspectral sensor, and include scenes from the Enrique Reef in LaParguera Puerto Rico. The models were then tested with real data. Results show that confidence intervals on the PCCcan be reliably obtained for real images.