Research On Hidden Targets Detection Algorithm Of Passive Millimeter Wave Imaging

The anti-terrorist environment of China has become more complicated,and we has been undertaking heavier responsibilities for anti-terrorism.As the support of antiterrorism forces,we must advance the security check technology with the times.Over the past decades,related researches indicate that passive millimeter wave imaging technology has become a trend in the field of security check,and hidden target detection algorithm has been one of the core technology in the passive millimeter wave imaging technology.Hidden target detection algorithm has had a lot of mature research and application case in the field of microwave and infrared imaging.As for the field of millimeter wave imaging,this technology belongs to the initial stage,and there are still a lot of issue and research to do.The research of this thesis bases on specific research projects,and it mainly includes:(1)On the basis of the passive millimeter wave imaging technology theory,the mathematical relationship,which is among the thermal radiation power of the target in the millimeter wave band and the self-temperature and the brightness temperature of the object,is established.And several millimeter wave image quality evaluation methods are researched.(2)Analyzing the causes of strip noise in imaging results of passive millimeter wave detection imaging system,a corresponding channel mismatch model is proposed.By doing the actual passive millimeter wave imaging experiment,the two-point channel calibration algorithm is evaluated,and we propose a improved algorithm for it.(3)Based on the sample data obtained from the system developed by the actual scientific research project,the corresponding simulation experiment is carried out by the two traditional multi-channel imaging image domain equalization algorithms,which are the histogram matching method and the Fourier transform method.And we analyze the performance of the two equalization algorithm from the experimental results to determine the research direction of image domain channel equalization in the field of millimeter wave image.The moments matching method is studied emphatically.Based on the statistical information of millimeter wave images,a multi-channel imaging adaptive equalization algorithm of pixel domain is researched and proposed.Also,we gain the outstanding experiment results.(4)The support vector machine(SVM)theory is studied.Under the condition of regression model of classifier,the problem of classifying hyperplane in classifier is analyzed and solved mathematically.On the basis of this,an improved hidden target detection algorithm based on support vector machine(SVM)is proposed by studying the scale invariant feature transform(SIFT)operator and the digital image connectivity domain.All the experimental data samples in this paper are derived from the actual research experiments,and the results verify the validity of the corresponding research contents.In summary,all the improved algorithms in this paper can effectively raise the detection performance of hidden targets in millimeter wave images,and they have practical theoretical reference and engineering application value for passive millimeter wave detection imaging technology.