Improving the spectral resolution for closely spaced targets based on MUSIC algorithm.

Multiple signal classification (MUSIC) is a high-resolution method in microwave imaging which is severely based on the number of singular values of a multistatic response matrix. However, the role of noise and multiple scattering is crucial to determine the true dimension of the signal subspace. In this article, we show that in the presence of multiple scattering between the targets, some singular values of the signal subspace will greatly be decreased which is a critical point when the noise is added to the system. In this case, these signal singular values will mix with the noise ones and the most popular MUSIC methods degrade and even fail to estimate closely spaced target locations. Here, we introduce a method to solve the ambiguity in determining the number of targets employing a pre-processing method, discrete stationary wavelet transform (DSWT). The DSWT is developed on the basis of features of the reflection coefficient between targets. The application of this algorithm to measurement data shows its superiority over another thresholding-based algorithm, empirical mode decomposition. The results indicate that DSWT+MUSIC yields accurate estimate of the target locations, even in the presence of considerable noise and multiple scattering in the received signal. [ABSTRACT FROM PUBLISHER]