Scattering Center Extraction for ISAR Image using Deep Neural Network

The scattering centers correspond to the reflective features of the target in the inverse synthetic aperture radar (ISAR) imagery. The accurate extraction is thus essential for target recognition. The CLEAN algorithm has been applied for scattering center extraction using the shooting and bouncing ray technique. Recently, we improved the above algorithm using the correlation of the point spread function. However, the CLEANbased algorithms still require an iterative peak search process. Therefore, we first apply the semantic segmentation based on the deep neural network to extract scattering centers. Numerical simulations for ideal point sources and contiguous target are performed to demonstrate the improvement of the proposed algorithm.