Micro-doppler-based Human Detection And Feature Extraction Algorithm

The movements of human body segments such as the heart's periodical beating, the periodical fluctuation of skin caused by breathing, the limbs'swing, etc. may induce non-linear phase modulations on the radar echo, which is called the micro-doppler effect. By the modulations, the human body's non-contact detection, location, imaging, recognition, behavior classifying and extraction of multiple human characteristics such as the frequencies of heartbeating and breathing, the gait frequency, etc. can be realized, and have plenty of applications and significance in the areas of military affairs, security system, anti-terrorism, rescue service, medicine monitoring, etc.In this paper, the issue of human detection and characteristic extraction is researched on two aspects of cardiopulmonary based micro-doppler and gait based micro-doppler. The main works are as follows:1. A gait echo model is proposed, which represents the instantaneous rotation angles or displacements of joints and their relative phase relationship, the instantaneous displacements and velocities of scatterers and their relative phase relationship, the instantaneous Radar Cross Section (RCS) and their relative phase relationship. The simulation result of the model and the experiment result match well, and the model provides the theoretical model foundation for human sensing and characteristic extraction.2. A clutter suppression-spectrum analysis detection algorithm of heartbeating and breathing for stepped-frequency continuous-wave is proposed. It has better computation efficiency, has little interference by noise and sidelobes, and never affect by the division of range cells compared with the current algorithm.3. The synthetic heterodyne demodulation is adopted to detect the heartbeating and breathing. There is no harmonic by this method which is fit for the condition of high signal to noise ratio.4. A human motion track estimation method based on Radon transformation is proposed, which is fit for the stepped-frequency continuous-wave human sensing and can estimate human rectilinear motion track. The relative theories and algorithms are verified by simulations and experiments. The results reveal that the methods based on cardiopulmonary micro-doppler and gait micro-doppler can realize the human detection and characteristic extraction.