Akademik Veri Yönetim Sistemi
IEEE Transactions on Instrumentation and Measurement, 2026 (SCI-Expanded, Scopus)
Radar-based microwave imaging systems (MIS) are well established, but enhancing image quality remains a challenge. This study proposes an approach that combines classical image processing with two novel steps. First, the collected S21 complex raw data are regularized using circular statistics to address phase interleaving issue inherent in angular phase data. Second, a multi-frequency diversity technique, known as image diversity, compares different images at the pixel level to select the most relevant data for each pixel, producing an improved reconstructed image. Experimental biomedical imaging results demonstrate accurate localization and dimensional estimation of a bone target embedded at a fixed depth within a moisture-controlled medium designed to mimic the dielectric properties of human tissue. The proposed approach is not limited to biomedical imaging, as its core innovations circular statistics for angular data and multi-frequency enhancement are broadly applicable to diverse sensing scenarios. For multi-frequency images, signal to clutter ratio (SCR), structural similarity index (SSIM), and cosine similarity reached 15 dB, 0.65, and 0.95, respectively. Implementing the frequency diversity technique improved these metrics by 27%, 59%, and 27%. The robustness of the reconstruction was evaluated by introducing varying levels of Gaussian noise into the raw data. Additionally, three widely used reconstruction methods Delay-and-Sum, E-Field, and linear phase–based approaches were implemented, and a comparative analysis table was provided.