报告题目: Dummy Variable Trick for Hyperspectral Anomaly Detection
报告人: Prof. Chein-I Chang, The University of Maryland, USA
时间：8月8号（周二），10:30-11:30
地点：江湾校区交叉二号楼B5007会议室
联系人: 王斌
摘要：Anomaly detection (AD) is quite different from target detection (TD) because it does not require target knowledge a priori as target detection does. As a result, AD is performed in passive mode as opposed to TD performed in active mode. Specifically, their applications are completely different in the sense that AD is mainly used for surveillance compared to TD used for reconnaissance. So, generally speaking, TD cannot be directly applied to AD. This talk presents a Dummy Variable Trick (DVT) theory which can convert TD to AD. In particular, we will show that the widely used anomaly detector developed Reed and Xiaoli, referred to as RXAD can be actually derived from likelihood ratio test (LRT) via DVT as generalized LRT (GLRT). Furthermore, by virtue of DVT, the constrained energy minimization (CEM) can be extended to CEMAD, which is exactly the commonly used RAD that uses correlation matrix to perform AD. What is more, the orthogonal subspace projection (OSP) can be also extended to OSPAD via DVT as a counterpart of unsupervised OSP target detector, referred to as automatic target generation process (ATGP). Consequently, AD can be considered as a dual theory of TD.
主讲人：Chein-I Chang received Ph.D. degree in electrical engineering from the University of Maryland, College Park and has been with the University of Maryland, Baltimore County (UMBC), since 1987 and is currently a Professor in the Department of Computer Science and Electrical Engineering. He is also a Chang Jiang Scholar Chair Professor and the director of Center for Hyperspectral Imaging in Remote Sensing (CHIRS) at Dalian Maritime University, Dalian, China since 2016.
Dr. Chang authored four books, Hyperspectral Imaging: Techniques for Spectral Detection and Classification published by Kluwer Academic Publishers in 2003 and Hyperspectral Data Processing: Algorithm Design and Analysis, John Wiley & Sons, 2013, Real Time Progressive Hyperspectral Image Processing: Endmember Finding and Anomaly Detection 2016 by Springer and Recursive Hyperspectral Sample and Band Processing: Algorithm Architecture and Implementation, Springer 2017 and a fifth forthcoming book, Hyperspectral Target and Anomaly Detection from Statistical Signal and Image Processing Perspectives, Wily, 2024. In addition, he also edited three books, Recent Advances in Hyperspectral Signal and Image Processing, 2006, Hyperspectral Data Exploitation: Theory and Applications, John Wiley & Sons, 2007 and Advances in Hyperspectral Image Processing Techniques, Wiley, 2023 and co-edited with A. Plaza a book on High Performance Computing in Remote Sensing, CRC Press, 2007.
Dr. Chang is a Life Fellow of IEEE and a Fellow of SPIE. He is currently an associate editor for Remote Sensing and IEEE Transaction on Geoscience and Remote Sensing.
