Currently, Doppler radars are widely used to detect and identify targets in different knowledge fields such as aerospace, medical or road safety. In this paper, a novel decision support system that relies on this kind of radars is presented. The aim of this system is to propose a time-frequency algorithm based on micro-Doppler signatures to detect and identify complex moving targets on public roads. Before obtaining this signature, each radar signal is improved using pre-processing techniques. Then, the short-time Fourier transform is applied to analyse and represent each signal received from the radar. Finally, different scenarios including pedestrians, vehicles and cyclists were used to demonstrate the performance of our system in target detection. The results have validated that pedestrians can be correctly identified only relying on their micro-Doppler signatures. As an example, this paper presents several use cases in which pedestrian signatures are characterized by vertical lines unlike vehicles and cyclists where signatures have horizontal lines characterized by larger vertical strips.