Research On Aeromagnetic Data Processing Based On Sparse Representation

A ferromagnetic target can produce an anomaly in the surrounding geomagnetic field. Detection of this anomaly in the geomagnetic field is usually accomplished by extracting he anomaly signal from the output of an airborne magnetometer, which is the so-called aeromagnetic survey. Aeromagnetic survey has a wide range of applications including mineral exploration, geological studies, etc. However, besides the anomaly signal caused by ferromagnetic target, the output signal of the airborne magnetometer contains the geomagnetic field, the interferences caused by the aircraft and other device noise. Moreover, since the magnetic anomaly signal is much weaker than the geomagnetic field and the aircraft interferences and their frequency ranges are overlapped, the traditional weak signal detection methods cannot be applied directly. Therefore, the geomagnetic field and the aircraft interferences should be removed before detecting the anomaly signal. In order to remove the aircraft interference, the socalled aeromagnetic compensation is necessary. Generally, the Tolles-Lawson model is used to describe the aircraft interference and its coefficients should be determined with high accuracy prior to the actual aeromagnetic survey, which yet is restricted by the poor constrained nature of the system of linear equations stablished for estimating the coefficients. Also, the geomagnetic field should be removed from the output of the airborne magnetometer before estimating the coefficients of the Tolles-Lawson model, which is often accomplished with a high-pass filter. However, he high-pass filter is incapable of eliminating the geomagnetic field thoroughly and yet may change the original waveform of the weak magnetic anomaly signal.Over the last few years, researches on signal sparse and redundant representations have made much achievements and the signal sparse decomposition model turns out to be an efficient framework in different fields. Considering the inefficiency of the high-pass filter in separating the geomagnetic field and the aircraft interference, as well as the singularity in estimating the coefficients of the Tolles-Lawson model, we try to incorporate the processing of the aeromagnetic measurement signal into the framework of signal sparse and redundant representations, and establish a delicate model for the decomposition, reconstruction, detection of the aeromagnetic measurement signal.The contributions of this thesis includes the following aspects:Firstly, we propose a MAD-KSVD dictionary learning method applicable to the reconstruction of the aeromagnetic measurement signal. By analyzing the varying characteristics of the geomagnetic field and the aircraft interference, the sparsity of the two different signals is determined according to the sparsityselecting theory. The proposed method takes advantages of the basis-pursuit method for selecting the sparsity and the KSVD method for updating the dictionary atoms. Simulation results demonstrate the efficiency of the proposed method in addressing the problems from the singularity in estimating the coefficients of the Tolles-Lawson model.Secondly, the morphological component analysis method is adopted to separate and compensate the aircraft interference. The global dictionary of the total magnetic field signal is built through analyzing the similarity between the atoms and the outliers in the reconstructed result. Then, the global dictionary is decomposed into two parts for the geomagnetic field and the aircraft interference, respectively, by using the modified MCA method and the eigenvectors corresponding to the atoms in the global dictionary. Consequently, the geomagnetic field and the aircraft interference are extracted from the measured total field signal based on the framework of MCA. Simulation results show that the magnitude of the residual interferences in the compensated signal obtained by the proposed method based on MCA is smaller than that obtained by the traditional method, illustrating the efficiency of the proposed method.Lastly, an aeromagnetic compensation system is designed and implemented. The system is able to acquire and save the scalar and vector magnetic field data from an RS232 serial port, and compensate the aircraft interference in real-time. Also, different methods have been packaged in the system, which is worthwhile for the research of aeromagnetic compensation.