Hidden Explosives Terahertz Detection,

Terahertz technology is a rising sensing and imaging technique over last years. The main work of this thesis contains two parts: firstly, point by point image experiments of explosive samples concealed by wrappers (letters, textiles and polythene) were performed by terahertz time domain spectroscopy system; secondly, image data were identified by Back Propagation artificial neural network (BP neural network). The experimental results show that explosive samples and non-explosive samples concealed by all kinds of wrappers can be highly accurately identified by BP neural network, although wrappers greatly influence the spectrums of samples. Moreover, identification precision of BP neural network can be advanced to a certain extent by optimizing spectrums using Burg algorithm. The work achieves some basic studies for the application of terahertz imaging technique and discusses its future in the security inspection field.Terahertz (THz, lTHz=10¹²Hz) radiation with a frequency range of 0.1THz¹0THz, sandwiched between microwave and infrared, belongs to far infrared region. In recent years, rapid progress in ultrafast laser technology provides a steady and available source for terahertz pulses generation, which greatly promotes the development of THz spectroscopy and imaging technique. The technique, as a new THz detection technique, has been widely used in the sensing and imaging field since it has the advantage of insensitive to thermal background and high signal-noise ratio comparing with the traditional Fourier transform infrared spectroscopy (FTIR).From 1970s, implementing security examination to people pass the important passageway (airport, port, station and custom) has became a common safety precaution all over the world. Over recent years, effecting by the queasy situation of the world, public security has became the focus of every government around the world. For keeping away and beating the occurrence and spread of crime, police of every country use new technology of security examination to check dangerous goods or illegal things. Terahertz technique can become an effective instrument of security examination likely due to its properties.Chapter one introduces the basic theories of THz spectroscopy and imaging technique, including the terahertz pulses generation, the terahertz pulses detection and its characteristics.The general devices of security check and the state of THz pulses imaging technique in the security check field were summarized. Our major work and signification for this thesis were presented.The principle of optical path which was built by our laboratory, experimental measurement procedure and sample preparation were presented in detail in chapter two. Moreover, the chapter especially emphasized on the principle part which were used in the article, including how to calculate optical constant of sample, the development of artificial neural network, the basic principle of BP neural network and Burg algorithm which was used for optimizing spectrums of samples.Chapter three illustrates the experimental results which eight samples (RDX, HMX, TNT, Salt, Sugar, Starch, Bezoar, Panax) respectively hidden by three types of letters (white letter, yellow letter, express mail letter) were identified by BP neural network. The absorption spectrums of hidden samples and naked samples were presented in the chapter. Comparing the difference between them, it should be noted that letters greatly influence the spectrums of samples. In the spectrums of hidden samples, some absorption peaks were attenuated and even disappeared, which were clear in the spectrums of naked samples. However, there are still faked absorptions peaks in places due to the reflection and scattering of terahertz pulses, which have not appeared in the spectrums of naked samples.The positions of these samples can be easily found in experimental images by identifying terahertz spectrums of hidden samples using BP neural network. Nevertheless, there were still some inaccurate identification points in images due to the influence of letters. Later, we optimized the spectrums of samples using Burg algorithm and sufficiently gain the optical information of them before identification. The accuracy of identification was enhanced greatly when the optimized data was inputted into the network. It demonstrates that using Burg algorithm greatly contributes to the action of BP neural network to highly accurate identification of explosive things hidden by letters.Chapter four illustrates the experimental results which eight samples respectively hidden by three types of textiles (cotton knitting fabric, dacron, artificial silk) were identified by BP neural network. The absorption spectrums of hidden samples were presented and the differences between these spectrums and the spectrums of naked samples were explained in the chapter.According to the data processing procedure which was used in chapter three, the spectrums of these samples were optimized by Burg algorithm and were identified by BP neural network. The results show that BP neural network has the good capability of identification to these samples hidden by textiles, and the capability could be strengthened using Burg algorithm.Chapter five illustrates the experimental results which eight samples respectively hidden by three types of polythene (dark plastic bag, polystyrene foam, hard plastic) were identified by BP neural network. According to the method which was used in the chapter three and chapter four, the positions of these samples can be well identified by BP network too. It demonstrates that artificial neural network could be used for highly accurately identifying explosive things hidden by all kinds of wrappers.In chapter six, we use other two methods of pattern recognition, including Euclid classifier and component analysis of spatial, to recognize these eight samples hidden by yellow letter. Comparing with artificial neural network, their advantages and disadvantages were presented in the chapter.Chapter seven is the summary of my work including experimental results and pattern recognition. The future and hurdles of terahertz imaging technique in the security inspection were discussed.