Speaker: Bai Xueru

Affiliation: Xidian University

Academic title: Professor

Abstract:

This lecture will introduce our recent progress in high-resolution ISAR imaging of aerospace targets within the framework of sparse Bayesian learning, which can effectively tackle the issues of low signal to noise ratio (SNR) and incomplete data observation by exploiting the statistical characteristics of both the target and the environment. Specifically, the probabilistic model and inference of the posterior will be introduced, together with a matrix inverse-free solution. Then, the corresponding deep network is constructed, which can facilitate the algorithm convergence and boost the imaging performance. Furthermore, to deal with the issue of repeated network training using sample sets with different SNRs, a hyper-network is designed. In particular, it exhibits robustness to various SNRs and needs to be trained only once. Finally, future work will be discussed.

Biography:

Dr. Bai is a full professor and doctoral advisor with the National Key Lab of Radar Signal Processing. Her research interests include high-resolution radar imaging and radar automatic target recognition. She was the recipient of Yong Science Award granted by the Ministry of Education, and received 1 Second Prize of the State Technological Invention Award and 2 ministerial or provincial–level prizes as the third contributor. She also received the Excellent Young Scientist Foundation granted by the National Natural Science Foundation of China and the National Program for Support of the Leading Innovative Talents.