Obtaining accurate real-time traffic information of highways, roads and bridges in urban and rural regions is important to intelligent transportation network management. On this thesis, an efficient method is proposed for improving the traffic flow detection using non-contact microwave radar detectors. Specifically, signal spectrums are analyzed for distinguishing the echo signals of a vehicle, which are reflected from frequency modulated continuous radar waves, from the background noise. Then, adequate detection thresholds are set to accurately identify the signal of a vehicle within a lane. The proposed scheme can enhance the detection accuracy of both lane vehicle flow and directional traffic volume. On-site road traffic experiments are performed to assess the proposed method. The results demonstrate the improvements of the proposed algorithm over the conventional scheme.