Anomaly Detection in Ground Object of Underground Nuclear Explosion Using TM Image

摘要

The comparison of multi-spectral images (so-called change detection) before and after the underground nuclear test can reduce search area for the on-site inspection and assist us in locating underground nuclear explosion (UNE). With TM images and multivariate alteration detection method, anomaly detection in ground object of a UNE was carried out by taking Pakistan UNE event as an example. In this paper a post-processing method was proposed based on the multivariate alteration detection. The correlation analysis between multivariate alteration detection variables and original multi-spectral images was studied as well as abnormal spectral characteristics of ground object associated with the UNE. The results indicated that the surface anomaly associated with the UNE partially corresponds to the observable phenomenon on the multi-spectral image and it is detectable through certain methods. The results also suggest that the multivariate alteration detection be in good generality on detecting anomaly in ground object of the UNE.