In this paper, a new hybrid classification method using both range profile (RP) and time-frequency image is proposed. The time-frequency image is obtained using the short-time Fourier transform before calculating the RP and this image is used for classification. 2-Dimensional Principal Components Analysis (2DPCA) is used to further compress the time-frequency image and to derive useful features from the image. The proposed method achieves a higher correct classification ratio than existing methods, especially when the signal-to-noise ratio is low.

Abstract-In the squint mode airborne spotlight synthetic aperture radar system using the range migration algorithm (RMA), autofocus (AF) technique yields poor results due to the squint spreading of the point spread function (PSF) of a scatterer. Thus, two-dimensional (2D) interpolation is required to direct PSF blurring in cross-range direction, to improve the cross-range resolution ∆y and to remove the spatially-varying sidelobe. Because conventional 2D interpolation requires huge computation time and yields large computation errors, we propose an efficient 2D interpolation technique for squint-mode RMA composed of two 1D interpolations. Simulation results using the measured turbulence data show ∆y was improved considerably and PSF was successfully focused by the proposed method with a reduced computation time.

Abstract-We propose an improved method of iterative physical optics (IPO) to analyze electromagnetic scattering by open-ended cavities.The traditional IPO method uses a fixed number of iterations; if this number is too small, the accuracy of the estimated monostatic radar cross section (RCS) of open-ended cavities degrades as the incident angle of the incident field increases. The recentlyintroduced adaptive iterative physical optics-change rate (AIPO-CR) method uses a variable number of iterations; compared to the IPO method, it predicts monostatic RCS more accurately, but requires more computation time. In this paper, a new algorithm is devised to improve both the monostatic RCS prediction accuracy of the IPO method, and the computational efficiency of the AIPO-CR method. The proposed method, iterative physical optics-retained previous solution (IPO-RPS), calculates the currents at one incident angle, then reuses them as the initial currents of iterations for the next incident angle. In simulations of the monostatic RCS for various open-ended cavities, the IPO-RPS method was more accurate than the traditional IPO method, and computationally more efficient than both the IPO and AIPO-CR methods.