Research On Human Detection Algorithm Based On Micro-perturbation And Gait Characteristics

Human detection based on microwave radar use of the Doppler effect from the human’s activity to realize the detection, location and tracking of human. It has important applications in modern urban war, natural disaster relief, medical care and other areas. The complexity of the scene to detect human body, the variety of human echo forms, and the faintness of the energy of human echo, make it difficult to effectively detect human. This paper focuses on the through walls human detection problem research, the main work and contributions are as follows:1. For the detection of human behind the wall, the model of human echo is studied, and echo signal from human is analyzed. Through simulation, determine the optimal band of transmitted radar signal under the cases of detecting human through wall;2. For the traditional human detection algorithm is failure in the condition of the scene which is non-Gaussian, a dual-channel correlation entropy detection algorithm is proposed. Firstly, by calculating the entropy of the echo signal this method achieves the accumulation of correlation energy and the suppression of clutter. Secondly, the human detection is achieved by making Fourier transform of the correlation entropy. This algorithm improves the SNR of the detection results effectively.3. The human micro-perturbation echo energy is weak and single-echo image contains less information, so a method using two images of the human body echo based on correlation analysis is proposed, through the correlation analysis between goals and scenarios, the energy of weak target signal is effectively strengthened, that help to extract human body’s weak information from the strong clutter;4. For the case that human gait Doppler echo signal contains a wealth of information which characteristic parameters need to be extracted from, the time-frequency analysis method is used to analysis human gait echo signal. Human-step echo frequency map is acquired through the short time Fourier transform. Summed up the method of extracting characteristic parameters of human gait;5. For the variety of human echo forms, support vector machine classification method is used for state recognition in the field of human movement.