Air-to-Ground Automatic Target Recognition and Estimation of Target Orientation Using Range Profiles

Automatic recognition of targets on the surface is of great importance, especially for active electronic scanning radar systems. The information estimated during the target recognition phase and the accuracy of the prediction directly affect the course of the operation. In this study, the performance of a radar system operating in the X frequency band in determining the classes and horizontal orientation angles of targets by using high resolution range profiles collected from different ground-based targets was examined. Examinations using the template matching technique together with deep learning-based sequential classifier structures show that the main target class can be recognized at levels above %85 depending on the signal-to-noise ratio. Depending on the symmetry in the structures of the targets, it has been observed that correct determination rate of horizontal orientation is generally divided into symmetrical orientation angle segments.