A Study for Hyperspectral Anomaly Change Detection on “Viareggio 2013 Trial” Dataset

Abstract

Hyperspectral anomaly change detection aims at finding rare and anomalous changes in multi-temporal hyperspectral images. There are existing many works about anomaly change detection algorithms, whereas they are all proposed and evaluated on their own datasets. With the publication of “Viareggio 2013 Trial”, it is necessary to compare the state-of-the-art methods on this dataset with fully ground-truth references. In this paper, we compare 8 anomaly change detection methods on the two multi-temporal pairs of “Viareggio 2013 Trial”. The experimental results indicate that slow feature analysis with LCRA obtains the best performance.