Antennas With High Radiation Efficiency And Objects Detection For Ground Penetrating Radar

The antenna and the detection algorithm are two key problems of impulse ground penetrating radar(GPR). In this thesis, these two crucial problems been both deeply studied, by which the probing ability of GPR can be explicitly improved.Before preparing to study the antenna, the impulse radiation is explained in time domain, which is the theory of antenna research in this thesis. The wideband nature of bow-tie antenna is very attractive for impulse radiation. And with its simple geometry and linear phase response, bow-tie antenna is prevalent in impulse GPR. Therefore, bow-tie antenna is selected to be researched. After the basic characters of bow-tie antenna been studied, a principle of choosing the geometry of bow-tie antenna and a principle of choosing the exciting pulse are proposed, which solves the basic design problem of bow-tie antenna.Because the antenna of impulse GPR works in time domain and near field, the classic characters in frequency domain and far field are not suitable to the antenna of impulse GPR. To solve this problem, the pulse radiation efficiency is defined. According to the definition, the pulse radiation efficiency means the ability of transforming the exciting pulse to the high quality probing pulse, which indicates the pulse has high main pulse and low late-time ring. The low radiation efficiency is the fatal problem of bow-tie antenna. In this thesis, the improvement of radiation efficiency is carried out by enhancing the main pulse and suppressing the late-time ring.In order to enhance the main pulse, an ultra-wide band balun is designed to increase the energy fed into an antenna. And after that, a new enhancing method by using the energy in late-time ring is proposed. The new method can increase the peak of main pulse as high as 100% comparing to a normal bow-tie antenna. Because the resistive loading is the most effective method of suppressing the late-time ring, the end loading and distribute loading are studied. For end loading, the problems of loading position, loading number and loading value are studied. As to solve the problem of enhancing main pulse in distribute loading, a loading method which can both suppressing the late-time ring and enhancing the main pulse is presented.Considering that the feature extraction is very crucial to object detection, the features of GPR objects are deeply studied before beginning to study the detection algorithm. And then a detection scheme is presented. After that, a detection algorithm based on the hidden Markov model(HMM) and an fast algorithm based on the cross correlation and Hough transform are proposed. Comparing to some existed detection methods, the HMM method has excellent performance but with the price of low computation efficiency. And the method based on the cross correlation and Hough transform can reach high computation efficiency without explicit decrement of detection performance. In order to decrease the false alarm, a method based on the interference theory is presented to extract the phase of object reflection, which can be used to discriminate objects according to the materials. Additionally, a method base on the Matching pursuit is proposed to extract the parameters of the GTD model, which can be used to discriminate objects from the geometrical shape.