Dual-View Hyperspectral Anomaly Detection via Spatial Consistency and Spectral Unmixing

Anomaly detection is a crucial task for hyperspectral image processing. Most popular methods detect anomalies at the pixel level, while few algorithms anomaly only utilize subpixel level unmixing technology to extract features without fundamentally analyzing anomalies. To better and separate from background, this paper proposes dual-view method by taking account of analysis both levels mentioned. At spectral angular distance adopted calculate similarities between central its neighbors in order further mine spatial consistency detection. On other hand, aspect analysis, it considered that difference background usually arises dissimilar endmembers, where will be fully implemented. Finally, results views are fused obtain Overall, proposed algorithm not interprets analyzes dual levels, but also employs Additionally, performance multiple data sets confirmed effectiveness algorithm.