Application of frequency correlation function to radar target detection

• Published in: IEEE transactions on aerospace and electronic systems 2003-01, Vol.39 (1), p.125-139
• Main Authors: El-Rouby, A.E., Nashashibi, A.Y., Ulaby, F.T.
• Format: Article
• Language: English
• Subjects: Algorithms; Detection algorithms; Feature based; Frequency; Image analysis; Mathematical analysis; Mathematical models; Object detection; Radar; Radar applications; Radar cross section; Radar detection; Radar imaging; Studies; Synthetic aperture radar; Trees; Trunks; Vehicles
• ISSN: 0018-9251; 1557-9603
• DOI: 10.1109/TAES.2003.1188898

Analysis of high-resolution 35 GHz synthetic aperture radar (SAR) imagery of terrain reveals that when point targets, such as vehicles, are viewed at angles close to grazing incidence, they are often difficult to distinguish from tree trunks because the radar cross section (RCS) intensities of the vehicles are comparable to the upper end of the RCS exhibited by tree trunks. To resolve the point target/tree trunk ambiguity problem, a detailed study was conducted to evaluate the use of new detection features based on the complex frequency correlation function (FCF). This paper presents an analytical examination of FCF and its physical meaning, the results of a numerical simulation study conducted to evaluate the performance of a detection algorithm that uses FCF, and the corroboration of theory with experimental observations conducted at 35 and 95 GHz. The FCF-based detection algorithm was found to correctly identify tree trunks as such in over 90% of the cases, while exhibiting a false alarm rate of only 3%.