Mr. Jamal Kazazi, Dr. Mahmoud Kamarei, Dr. Mohammad Fakharzadeh,
Volume 15, Issue 4 (12-2025)
Abstract
Target detection using cameras or automotive radar to identify traffic or prevent collisions is an important issue in Autonomous Vehicles (AV) research. Traditional Constant False Alarm Rate (CFAR) methods are commonly employed. Although these methods are suitable for lightweight hardware, improving the target detection process often leads to losing real-time performance. The method proposed in this paper improves detection accuracy. It reduces response time by modifying the position of guard cells in the first stage and employing harmonic averaging (inverse of the sum of the inverse of data) while eliminating data sorting in the second stage. Moreover, this approach exhibits better performance in the presence of interfering targets. Since the proposed method is more applicable to the Range-Doppler map, it has been named RD-CFAR. The proposed method also enhances target detection in Synthetic Aperture Radar (SAR) images. Simulation results demonstrate that the proposed algorithm improves detection probability by nearly 40% compared to conventional methods (like CA-CFAR), while maintaining comparable computational time.
