Multi-sensor anomalous change detection at scale

摘要

Combining multiple satellite remote sensing sources provides a far richer, more frequent view of the earththan that of any single source; the challenge is in distilling these petabytes of heterogeneous sensor imageryinto meaningful characterizations of the imaged areas. To meet this challenge requires effective algorithms forcombining heterogeneous data to identify subtle but important changes among the intrinsic data variation. Themajor obstacle to using heterogeneous satellite data to monitor anomalous changes across time is this: subtle butreal changes on the ground can be overwhelmed by artifacts that are simply due to the change in modality. Here,we implement a joint-distribution framework for anomalous change detection that can effectively "normalize"for these changes in modality, and does not require any phenomenological resampling of the pixel signal. Thisexibility enables the use of satellite imagery from di erent sensor platforms and modalities. We use multi-yearconstruction of the Los Angeles Stadium at Hollywood Park (in Inglewood, CA) as our testbed, and exploitsynthetic aperture radar (SAR) imagery from Sentinel-1 and multispectral imagery from both Sentinel-2 andLandsat 8. We explore results for anomalous change detection between Sentinel-2 and Landsat 8 over time, andalso show results for anomalous change detection between Sentinel-1 SAR imagery and Sentinel-2 multispectralimagery.