利用能量色散X射线探测复杂背景下毒品/危险品的新方法研究与应用
|
摘要
毒品和和恐怖活动是当今世界的两大毒瘤,严重危害着人类安全和健康,虽然世界各国都在严厉打击涉毒和恐怖等严重危害社会的犯罪活动,但是魔高一尺道高一丈,犯罪分子的作案手段也在不断的更新,打击涉毒和恐怖犯罪已成为当今世界各国政府和国际组织的共同目标,而要解决恐怖主义和毒品犯罪首先要解决毒品／爆炸物的现场快速检测。在公共安全场所的实际检测中,除了要检测常规的海洛因、可卡因、大麻等毒品,还需要检测能制造毒品的原料,即易制毒化学品。而爆炸物的检测除了常规的TNT等常规炸药,液体炸药和燃爆物由于组分复杂、性质均一、原料来源方便和加工工艺简单等原因,对公共安全造成了更大的威胁而迫切需要现场快速检测技术和仪器,因此本项目需要研究的检测目标包括毒品、爆炸物和液体危险品。
纵观国内外毒品、爆炸物和液体危险品的主要探测技术和方法包括：试纸和化学试剂法、电磁法、光学和光谱法、离子迁移谱法、色谱质谱联用法、超声法以及X射线探测等。上述毒品／爆炸物检测技术和方法主要有缺如下：试纸和化学试剂法优点在于操作简单,无需电力或电池驱动,但该方法属于有损探测,而且一种试纸或试剂只能探测一种特定的物质；电磁法主要有核磁共振、太赫兹、微波等探测方法,核磁共振探测的缺点是探测时间长且无法探测不含有元素的液体炸药,太赫兹的探测精度受环境影响较大,电磁探测方法共同缺点都是不能探测被金属包裹或密封的液体；光学探测方法主要有拉曼和红外光谱探测仪器,光学探测仪器属于无损检测,但是光学仪器只能检测透明包装材料中的物质,而且光路中的传播路径易受其他物质干扰而导致测量误差；离子迁移谱属于有损探测,而且本质上不能检查密闭容器内部的液体,同时该方法电子稳定性不高,还需要根据不同测量环境进行标定而影响精度,因此误检率高；色谱质谱联用探测方法的优点是探测精度高,但是缺点是探测过程复杂、速度慢、探测费用高以及属于有损探测等；超声探测的优点是测试仪器轻便,操作安全简单,缺点是必须预先测量容器尺寸,并且探测精度受到回波影响较大。
相对于上述毒品／爆炸物探测方法,X射线类探测方法的优点是非破坏性、非接触、分辨率高、低成本、穿透性强、低剂量对人体基本无害,因此国内外公共安全检查仪器基本上以X射线探测方法为主。X射线探测技术主要是通过物质的密度、图像或者原子序数来鉴别,虽然国内外采用X射线探测毒品/爆炸物取得了一定进展,但是还存在一定的局限性：X射线探测技术仅能获得的物质形状、密度或者原子序数,无法分辨出密度相近的物质；仅能获得液体物质一维或者二维信息,因此无法区分不同浓度或者混合的多组分液体；探测速度较慢,难以适合公共安全领域现场快速探测的要求。
近年来,随着X射线理论与探测技术的进步,能量色散X射线(Energy dispersive X-ray scattering)由于使用宽波段全波长扫描的方式,因此具具有探测速度快、操作方便、稳定性、重现性好、不必聚焦以及受环境影响较小等优点,成为固体晶体或多晶粉末衍射分析的主要谱仪。但是该类型探测仪器具有分辨率低的缺点,因此国内外少有用于毒品/爆炸物的现场快速检测。
本文使用能量色散x射线谱仪作为探测工具,针对复杂背景下的毒品/爆炸物探测需求,研究适用于现场便携快速检测新方法及实用化仪器的关键软硬件技术。研究的方向一是针对人体藏匿毒品的检测,二是针对液体炸药、易制毒和易燃爆等液体危险品的检测。虽然能量散射×射线探测仪器有不少优点,但是能量散射x射线属于弱光射线,分辨率低。而在人体藏匿毒品条件下,由于人体组织对光子的衰减作用和人体组织的复杂结构,导致能量散射x射线信号弱,分辨率更低,通过传统的寻找指纹峰的方法很难准确识别待测可疑物；另外,常规的x射线探测液体危险品主要是通过液体的密度或者原子序数来鉴别的,一般只能检测单组份液体危险品及原料,多组分液体危险品很难通过二维的谱识别方法来检测和识别。为了解决复杂背景下能量色散x射线的探测信号弱,难以直接通过寻峰方法识别待测物质的难题,采用智能算法辅助识别x射线谱是必要的,本文的具体研究方法和路线如下：
(1)设计和研制了两种能量色散x射线探测装备,分别是低能探测试验平台和高能多功能散射试验平台,通过在探测平台上研究弱X光在人体环境中探测毒品/爆炸物的衍射识别规律,探索人体隐藏毒品/爆炸物最佳检测条件；为研究弱X光能量色散探测毒品/爆炸物的新型检测原理和方法提供理论基础,为能量色散X射线探测数据的处理和识别提供可靠的数据。
(2)利用自主研制的能量色散X射线谱仪,本项目研究了能量色散X射线探测人体藏匿毒品及其智能识别算法,能量色散X射线属于弱X射线光束,因此在X光源能量相近的情况下,能散X射线探测的能量分辨能力要比常规小角散射X射线要低。另外,本项目的探测目标是人体藏匿等复杂背景下毒品／爆炸物探测,经过背景对光束的衰减作用,待测物质的散射信号比较弱,很难通过物质的衍射峰来鉴别物质类型。因此,本项目针对复杂背景下能量散射X射线数据,采用多种信号处理、模式分类、统计分析和智能识别等算法来处理获取的能量色散数据。通过阈值小波与相干函数组合算法,对不同尺寸和类型的模拟人体组织(猪肉组织)包裹海洛因的能量散射X射线数据进行处理,通过数据的去噪和相干函数匹配,建立模拟人体包裹海洛因能散数X射线探测数据库,然后进行识别；为了研究模拟人体组织包裹下不同种类毒品及相似物质的识别率,本项目选用模拟人体组织(猪肉)包裹TNT、苯乙酸、冰毒、海洛因、胡椒醛、人体和盐等7类物质作为测试物质,利用PCA算法来分类识别7中物质,通过神经网络来优化PCA算法,并研究了PCA算法中提取的主成分数目对识别率的影响规律；本项目还选用能量色散X射线探测医学人体模型藏匿毒品/爆炸物,然后采用不同的模式识别和统计算法来处理和识别待测物,待测物质为TNT,苯乙酸,冰毒,海洛因,胡椒醛,人体,盐,提出了一种基于粒子群算法优化的SVM分类识别算法,通过对每种物质30组数据的识别和优化,寻求最优性能的SVM识别能量色散X射线探测人体模型藏匿毒品/爆炸物识别算法；另外本项目还研究了基于离散余弦转换和线性判别组合算法对人体模型藏匿毒品/爆炸物识别算法,利用上述算法对高维数据的去相关性能量、特征提取和分类优势,解决弱X光数据的难以处理和识别率低的难点。最后,本章研究了边缘fisher分析对人体模型藏匿毒品/爆炸物识别,特别是边缘fisher分析对样本分布不作任何要求,而是利用邻近样本之间的几何关系表征样本分布的特性,能较好的解决弱X光数据的识别和分类问题,并对比和评价K-近邻法(KNN)和支持向量机(SVM)与PCA、PCA+LDA和MFA算法组合对提高人体模型藏匿TNT,苯乙酸,冰毒,海洛因,胡椒醛,人体,盐的识别效果,验证本文提出的算法有效性和高识别率。
研究液体危险品在能量散射X射线探测下的光学散射特性,选择适合液体危险品的最佳测试时间、电压、测试角度、狭缝宽度等,测试常见液体危险品样品,获取其三维能散数据。构建液体危险品能散X射线三维数据模型,对三维模型中各载荷矩阵多变量数据进行回归拟合和矫正,建立液体危险品能量色散X射线三维数据的PARAFAC模型,然后计算主因子数,接着对三维数据模型进行二维载荷矩阵分解,通过对二维载荷矩阵的分类和识别,最终计算出多组分液体危险品中各组分的精确成分和含量。
最后,本项目采用能量散射X射线探测方法,针对复杂背景条件下(人体或者人体模型)的毒品/爆炸物的能散X光谱特征,分别研究了统计模型、模式分类和识别算法、矩阵分析等职能算法,在对常见毒品/爆炸物和液体危险品分析和检测的基础上,提出了多种有效可行的探测方法,并有望为研制新型能量色散X射线探测仪器提供技术支持。
Drugs and terrorism are two big tumors in today's world, which cause serious damage to the human safety and health. As countries all over the world are striking sever blows to drug-related and terrorist criminal activities which seriously endanger the society, criminals are unceasingly updating their means of committing crimes. To fight against drug-related and terrorist crimes has become a common goal of the world's governments and international organizations. To achieve this goal we must first solve the problem of on-site rapid detection of drugs and explosives. During actual detection in places of public security, in addition to detecting the conventional drugs such as heroin, cocaine, cannabis, we also need to detect the raw material of drugs namely the precursor chemicals. Meanwhile, we have managed to detect the conventional explosives such as TNT. However, we urgently need technology and equipment to rapidly on-site detect liquid explosives and bums which cause more of a threat to public safety due to their complex composition, similar quality, convenient sources of raw materials and simple processing. So the project is mainly studying the detection of drugs, explosives and liquid dangerous materials.
Main techniques and methods of detecting drugs, explosives and liquid dangerous materials home and abroad include test paper and chemical reagent method, electromagnetic method, optical method and spectroscopy, ion mobility spectrometry, chromatography and mass spectrometry method, ultrasonic method X-ray detection, etc. All these detection technology and methods have advantages and disadvantages. Test paper and chemical reagent method is easy operated and dispenses with electricity or battery drive, but it belongs to destructive testing, and one paper or reagent can only detect one certain material. Electromagnetic methods mainly include nuclear magnetic resonance (NMR), terahertz, microwave detection method and so on. The disadvantage of nuclear magnetic resonance (NMR) is long detection process and inability to detect liquid explosives containing no element. The detecting precision of the terahertz is greatly influenced by the environment. The common defects of electromagnetic detection methods are inability to detect liquid wrapped or sealed by metal. Optical detection methods are mainly using Raman and infrared spectrum detection instrument. Optical instrument testing is nondestructive testing, but it can only detect materials with transparent packaging and optical propagation path susceptible to other substances'interference may lead to measuring error. Ion migration spectrum method is destructive testing and unable to detect liquid in an airtight container essentially. Meanwhile, its low electronic stability and measuring accuracy easily influenced by different calibrations according to different measurement environments result in high error detection rate. Chromatography and mass spectrometry detection method has the advantage of high accuracy, but it also has disadvantages of complex detection process, slow speed, high cost and destructive testing, etc. Ultrasonic detection has advantages of the portable test instrument, safe and easy operation, but you have to measure the container size in advance and the detection precision is largely affected by echo.
Compared with the drug/explosive detection methods mentioned above, X-ray detection method has advantages of non-destructive testing, non-contact testing, high resolution, low cost, strong penetrability and basically harmless to human body with low dose. Therefore, the domestic and foreign public security inspection instruments are basically given priority to X-ray detection method. X-ray detection technology is mainly identified by means of the density of material, images, and atomic number. While using X-ray to detect drug/explosive at home and abroad has made some progress, there are still some limitations:X ray detection technique can only obtain the material's shape, density and the atomic number, but unable to discern materials with similar density. It can only obtain liquid material's one-dimensional or two-dimensional information, but unable to distinguish liquids of different concentrations or mixed multi-component liquid. Moreover, its detecting speed is slow, which makes it difficult to meet the requirements of on-site rapid detection in the field of public safety.
In recent years, with the progress of X ray theory and detection technology, energy dispersive X-ray scattering with broadband and wavelength scanning becomes the main spectrometer of solid crystalline or polycrystalline powder diffraction analysis due to its fast detection speed, easy operation, stability, good reproducibility, no need of focusing, less environment influence. But this type of instrument is rarely used in on-site rapid detection of drug/explosives at home and abroad because of its low resolution.
By using energy dispersive X-ray spectrometer as its detection tool, this paper studies the new on-site portable rapid detecting method and key hardware and software technologies of practical instrument for the demand of the drug/explosive detection under complicated backgrounds. There are two research directions in this paper:detecting drugs hiding inside of human body and liquid dangerous materials such as liquid explosives, precursor chemicals and bums, etc. Although the energy dispersive X-ray detection instrument has many advantages, it belongs to weak light ray and has low resolution. When drugs are hidden inside of human body, energy scattering X-ray signal becomes weaker, and its resolution lower because of the body tissues'photon attenuation function and its complex structure, which makes it very difficult to accurately discern suspicious objects under test through traditional methods of looking for fingerprint peaks. In addition, the conventional X-ray method of detecting liquid dangerous materials is mainly through liquid density or the atomic number which normally only detect one-component liquid dangerous materials and raw materials while multi-component liquid dangerous materials are difficult to detect and identify through two-dimensional spectrum identification method. In order to solve this problem, it is necessary to apply intelligent algorithm to aide X-ray spectrum identification. The concrete research methods and routes of this paper are as follows:
(1) Design and develop two kinds of energy dispersive X-ray detection equipments, respectively low energy detection experiment platform and high-energy multi-functional scattering experiment platform; explore the best condition of detecting drugs/explosives hidden in human body through studying the X-ray diffraction recognition law by detecting drugs/explosives in the human environment on detection platform; provide theoretical foundation for studying new detecting principles and methods of using X-ray energy dispersion to detect drugs/explosives; provide reliable data for the data processing and recognition of energy dispersive X-ray detection.
(2) By using the energy dispersive X-ray spectrometer, this project studies the energy dispersive X-ray detection of drugs hidden in human body and its intelligent identification algorithm. Energy dispersive X-ray belongs to the weak X-ray beam, so energy dispersive X-ray detection's energy resolution is lower than conventional small angle scattering X-ray in similar X light source energy case. Besides, the detection target of this project is drugs/explosives hidden in human body. It is very difficult to identify materials through materials'diffraction peak when the scattered signal of the material under test is rather weak after background causing attenuation effect on the light beam.
Therefore, this project applies signal processing, pattern classification, statistical analysis and intelligent recognition and other algorithms to deal with the energy dispersive data obtained under complicated backgrounds. Using combinational algorithm of wavelet threshold and coherence function, we process the energy scattering X-ray data of heroin wrapped by different size and type of simulated human tissue (pork tissue). Through data denoising and coherence function matching, we establish simulated human tissue wrapped heroin's energy scattering X-ray detection database, and then identify it. In order to study the identification rate of different drugs and similar matter wrapped by simulated human tissue, this project chooses seven materials wrapped by simulated human tissue (pork tissue) including TNT, phenylacetic acid, methamphetamine, heroin, piperonal, human body and salt as test materials, uses PCA algorithm to identify them, optimizes PCA algorithm through the neural network and studies the influence law of the principal component number extracted from PCA algorithm on the recognition rate;
This project also chooses energy dispersive X-ray to detect drugs/explosives hidden in medical human body model, then adopts different pattern recognitions and statistical algorithms to process and recognize materials under test including TNT, phenylacetic acid, methamphetamine, heroin, piperonal, human body, and salt. The project puts forward a SVM classification algorithm based on particle swarm optimization. By the recognition and optimization of30sets of data of every material, we are seeking the optimal performance recognition algorithm of the SVM identification energy dispersive X-ray detecting drugs/explosives hidden in human body model. Furthermore, this project also studies a recognition algorithm of detecting drugs/explosives hidden in human body based on the discrete cosine transform and linear discriminant algorithm, and solves the problem of processing difficulty and low recognition of weak X-ray data by using this algorithm's advantage of extraction and classification of high dimensional data's decorrelation energy and feature.
At the end of this chapter, we studies the edge fisher analysis detecting drugs/explosives hidden in human body, especially without any demand on sample distribution, uses geometric relationships between the adjacent samples to represent the properties of the sample distribution, better solves the identification and classification problem of weak X-ray data, and applies the method of K-near neighbor (KNN), support vector machine (SVM), combined algorithm of PCA, PCA+LDA and MFA to raise the recognition effect of TNT, phenylacetic acid, methamphetamine, heroin, piperonal, human body, salt hidden in human body.
We research liquid dangerous materials'optical scattering characteristics under the detection of energy dispersive X-ray, choose the best test time, voltage, test angle, slit width etc. suitable for liquid dangerous materials, test common hazardous liquid sample, and get its3dimensional energy dispersive data. We build three-dimensional data model of liquid dangerous materials'energy-dispersive X-ray, regress, match and correct the load matrix's multivariate data in the three-dimensional model, establish three-dimensional PARAFAC data model of liquid dangerous materials' energy-dispersive X-ray, calculate the principal divisor number, then do the two-dimensional load matrix decomposition on the three-dimensional data model, classify and recognize the two-dimensional load matrix, and finally work out the accurate component and content of each material in the multi-component liquid explosive.
To sum up, in view of the energy dispersive X-ray spectrum characteristics of drugs/explosives in (human body or human body model) under the complex background condition, this project adopts the energy dispersive X-ray detection method to respectively studies the statistical model, pattern classification, recognition algorithm, matrix analysis and other function algorithm. On the basis of analysis and detection of common drugs/explosives and liquid dangerous materials, the project puts forward a variety of effective detection methods which may hopefully provide technical support for developing new pattern energy dispersive X-ray detection instruments.
纵观国内外毒品、爆炸物和液体危险品的主要探测技术和方法包括：试纸和化学试剂法、电磁法、光学和光谱法、离子迁移谱法、色谱质谱联用法、超声法以及X射线探测等。上述毒品／爆炸物检测技术和方法主要有缺如下：试纸和化学试剂法优点在于操作简单,无需电力或电池驱动,但该方法属于有损探测,而且一种试纸或试剂只能探测一种特定的物质；电磁法主要有核磁共振、太赫兹、微波等探测方法,核磁共振探测的缺点是探测时间长且无法探测不含有元素的液体炸药,太赫兹的探测精度受环境影响较大,电磁探测方法共同缺点都是不能探测被金属包裹或密封的液体；光学探测方法主要有拉曼和红外光谱探测仪器,光学探测仪器属于无损检测,但是光学仪器只能检测透明包装材料中的物质,而且光路中的传播路径易受其他物质干扰而导致测量误差；离子迁移谱属于有损探测,而且本质上不能检查密闭容器内部的液体,同时该方法电子稳定性不高,还需要根据不同测量环境进行标定而影响精度,因此误检率高；色谱质谱联用探测方法的优点是探测精度高,但是缺点是探测过程复杂、速度慢、探测费用高以及属于有损探测等；超声探测的优点是测试仪器轻便,操作安全简单,缺点是必须预先测量容器尺寸,并且探测精度受到回波影响较大。
相对于上述毒品／爆炸物探测方法,X射线类探测方法的优点是非破坏性、非接触、分辨率高、低成本、穿透性强、低剂量对人体基本无害,因此国内外公共安全检查仪器基本上以X射线探测方法为主。X射线探测技术主要是通过物质的密度、图像或者原子序数来鉴别,虽然国内外采用X射线探测毒品/爆炸物取得了一定进展,但是还存在一定的局限性：X射线探测技术仅能获得的物质形状、密度或者原子序数,无法分辨出密度相近的物质；仅能获得液体物质一维或者二维信息,因此无法区分不同浓度或者混合的多组分液体；探测速度较慢,难以适合公共安全领域现场快速探测的要求。
近年来,随着X射线理论与探测技术的进步,能量色散X射线(Energy dispersive X-ray scattering)由于使用宽波段全波长扫描的方式,因此具具有探测速度快、操作方便、稳定性、重现性好、不必聚焦以及受环境影响较小等优点,成为固体晶体或多晶粉末衍射分析的主要谱仪。但是该类型探测仪器具有分辨率低的缺点,因此国内外少有用于毒品/爆炸物的现场快速检测。
本文使用能量色散x射线谱仪作为探测工具,针对复杂背景下的毒品/爆炸物探测需求,研究适用于现场便携快速检测新方法及实用化仪器的关键软硬件技术。研究的方向一是针对人体藏匿毒品的检测,二是针对液体炸药、易制毒和易燃爆等液体危险品的检测。虽然能量散射×射线探测仪器有不少优点,但是能量散射x射线属于弱光射线,分辨率低。而在人体藏匿毒品条件下,由于人体组织对光子的衰减作用和人体组织的复杂结构,导致能量散射x射线信号弱,分辨率更低,通过传统的寻找指纹峰的方法很难准确识别待测可疑物；另外,常规的x射线探测液体危险品主要是通过液体的密度或者原子序数来鉴别的,一般只能检测单组份液体危险品及原料,多组分液体危险品很难通过二维的谱识别方法来检测和识别。为了解决复杂背景下能量色散x射线的探测信号弱,难以直接通过寻峰方法识别待测物质的难题,采用智能算法辅助识别x射线谱是必要的,本文的具体研究方法和路线如下：
(1)设计和研制了两种能量色散x射线探测装备,分别是低能探测试验平台和高能多功能散射试验平台,通过在探测平台上研究弱X光在人体环境中探测毒品/爆炸物的衍射识别规律,探索人体隐藏毒品/爆炸物最佳检测条件；为研究弱X光能量色散探测毒品/爆炸物的新型检测原理和方法提供理论基础,为能量色散X射线探测数据的处理和识别提供可靠的数据。
(2)利用自主研制的能量色散X射线谱仪,本项目研究了能量色散X射线探测人体藏匿毒品及其智能识别算法,能量色散X射线属于弱X射线光束,因此在X光源能量相近的情况下,能散X射线探测的能量分辨能力要比常规小角散射X射线要低。另外,本项目的探测目标是人体藏匿等复杂背景下毒品／爆炸物探测,经过背景对光束的衰减作用,待测物质的散射信号比较弱,很难通过物质的衍射峰来鉴别物质类型。因此,本项目针对复杂背景下能量散射X射线数据,采用多种信号处理、模式分类、统计分析和智能识别等算法来处理获取的能量色散数据。通过阈值小波与相干函数组合算法,对不同尺寸和类型的模拟人体组织(猪肉组织)包裹海洛因的能量散射X射线数据进行处理,通过数据的去噪和相干函数匹配,建立模拟人体包裹海洛因能散数X射线探测数据库,然后进行识别；为了研究模拟人体组织包裹下不同种类毒品及相似物质的识别率,本项目选用模拟人体组织(猪肉)包裹TNT、苯乙酸、冰毒、海洛因、胡椒醛、人体和盐等7类物质作为测试物质,利用PCA算法来分类识别7中物质,通过神经网络来优化PCA算法,并研究了PCA算法中提取的主成分数目对识别率的影响规律；本项目还选用能量色散X射线探测医学人体模型藏匿毒品/爆炸物,然后采用不同的模式识别和统计算法来处理和识别待测物,待测物质为TNT,苯乙酸,冰毒,海洛因,胡椒醛,人体,盐,提出了一种基于粒子群算法优化的SVM分类识别算法,通过对每种物质30组数据的识别和优化,寻求最优性能的SVM识别能量色散X射线探测人体模型藏匿毒品/爆炸物识别算法；另外本项目还研究了基于离散余弦转换和线性判别组合算法对人体模型藏匿毒品/爆炸物识别算法,利用上述算法对高维数据的去相关性能量、特征提取和分类优势,解决弱X光数据的难以处理和识别率低的难点。最后,本章研究了边缘fisher分析对人体模型藏匿毒品/爆炸物识别,特别是边缘fisher分析对样本分布不作任何要求,而是利用邻近样本之间的几何关系表征样本分布的特性,能较好的解决弱X光数据的识别和分类问题,并对比和评价K-近邻法(KNN)和支持向量机(SVM)与PCA、PCA+LDA和MFA算法组合对提高人体模型藏匿TNT,苯乙酸,冰毒,海洛因,胡椒醛,人体,盐的识别效果,验证本文提出的算法有效性和高识别率。
研究液体危险品在能量散射X射线探测下的光学散射特性,选择适合液体危险品的最佳测试时间、电压、测试角度、狭缝宽度等,测试常见液体危险品样品,获取其三维能散数据。构建液体危险品能散X射线三维数据模型,对三维模型中各载荷矩阵多变量数据进行回归拟合和矫正,建立液体危险品能量色散X射线三维数据的PARAFAC模型,然后计算主因子数,接着对三维数据模型进行二维载荷矩阵分解,通过对二维载荷矩阵的分类和识别,最终计算出多组分液体危险品中各组分的精确成分和含量。
最后,本项目采用能量散射X射线探测方法,针对复杂背景条件下(人体或者人体模型)的毒品/爆炸物的能散X光谱特征,分别研究了统计模型、模式分类和识别算法、矩阵分析等职能算法,在对常见毒品/爆炸物和液体危险品分析和检测的基础上,提出了多种有效可行的探测方法,并有望为研制新型能量色散X射线探测仪器提供技术支持。
Drugs and terrorism are two big tumors in today's world, which cause serious damage to the human safety and health. As countries all over the world are striking sever blows to drug-related and terrorist criminal activities which seriously endanger the society, criminals are unceasingly updating their means of committing crimes. To fight against drug-related and terrorist crimes has become a common goal of the world's governments and international organizations. To achieve this goal we must first solve the problem of on-site rapid detection of drugs and explosives. During actual detection in places of public security, in addition to detecting the conventional drugs such as heroin, cocaine, cannabis, we also need to detect the raw material of drugs namely the precursor chemicals. Meanwhile, we have managed to detect the conventional explosives such as TNT. However, we urgently need technology and equipment to rapidly on-site detect liquid explosives and bums which cause more of a threat to public safety due to their complex composition, similar quality, convenient sources of raw materials and simple processing. So the project is mainly studying the detection of drugs, explosives and liquid dangerous materials.
Main techniques and methods of detecting drugs, explosives and liquid dangerous materials home and abroad include test paper and chemical reagent method, electromagnetic method, optical method and spectroscopy, ion mobility spectrometry, chromatography and mass spectrometry method, ultrasonic method X-ray detection, etc. All these detection technology and methods have advantages and disadvantages. Test paper and chemical reagent method is easy operated and dispenses with electricity or battery drive, but it belongs to destructive testing, and one paper or reagent can only detect one certain material. Electromagnetic methods mainly include nuclear magnetic resonance (NMR), terahertz, microwave detection method and so on. The disadvantage of nuclear magnetic resonance (NMR) is long detection process and inability to detect liquid explosives containing no element. The detecting precision of the terahertz is greatly influenced by the environment. The common defects of electromagnetic detection methods are inability to detect liquid wrapped or sealed by metal. Optical detection methods are mainly using Raman and infrared spectrum detection instrument. Optical instrument testing is nondestructive testing, but it can only detect materials with transparent packaging and optical propagation path susceptible to other substances'interference may lead to measuring error. Ion migration spectrum method is destructive testing and unable to detect liquid in an airtight container essentially. Meanwhile, its low electronic stability and measuring accuracy easily influenced by different calibrations according to different measurement environments result in high error detection rate. Chromatography and mass spectrometry detection method has the advantage of high accuracy, but it also has disadvantages of complex detection process, slow speed, high cost and destructive testing, etc. Ultrasonic detection has advantages of the portable test instrument, safe and easy operation, but you have to measure the container size in advance and the detection precision is largely affected by echo.
Compared with the drug/explosive detection methods mentioned above, X-ray detection method has advantages of non-destructive testing, non-contact testing, high resolution, low cost, strong penetrability and basically harmless to human body with low dose. Therefore, the domestic and foreign public security inspection instruments are basically given priority to X-ray detection method. X-ray detection technology is mainly identified by means of the density of material, images, and atomic number. While using X-ray to detect drug/explosive at home and abroad has made some progress, there are still some limitations:X ray detection technique can only obtain the material's shape, density and the atomic number, but unable to discern materials with similar density. It can only obtain liquid material's one-dimensional or two-dimensional information, but unable to distinguish liquids of different concentrations or mixed multi-component liquid. Moreover, its detecting speed is slow, which makes it difficult to meet the requirements of on-site rapid detection in the field of public safety.
In recent years, with the progress of X ray theory and detection technology, energy dispersive X-ray scattering with broadband and wavelength scanning becomes the main spectrometer of solid crystalline or polycrystalline powder diffraction analysis due to its fast detection speed, easy operation, stability, good reproducibility, no need of focusing, less environment influence. But this type of instrument is rarely used in on-site rapid detection of drug/explosives at home and abroad because of its low resolution.
By using energy dispersive X-ray spectrometer as its detection tool, this paper studies the new on-site portable rapid detecting method and key hardware and software technologies of practical instrument for the demand of the drug/explosive detection under complicated backgrounds. There are two research directions in this paper:detecting drugs hiding inside of human body and liquid dangerous materials such as liquid explosives, precursor chemicals and bums, etc. Although the energy dispersive X-ray detection instrument has many advantages, it belongs to weak light ray and has low resolution. When drugs are hidden inside of human body, energy scattering X-ray signal becomes weaker, and its resolution lower because of the body tissues'photon attenuation function and its complex structure, which makes it very difficult to accurately discern suspicious objects under test through traditional methods of looking for fingerprint peaks. In addition, the conventional X-ray method of detecting liquid dangerous materials is mainly through liquid density or the atomic number which normally only detect one-component liquid dangerous materials and raw materials while multi-component liquid dangerous materials are difficult to detect and identify through two-dimensional spectrum identification method. In order to solve this problem, it is necessary to apply intelligent algorithm to aide X-ray spectrum identification. The concrete research methods and routes of this paper are as follows:
(1) Design and develop two kinds of energy dispersive X-ray detection equipments, respectively low energy detection experiment platform and high-energy multi-functional scattering experiment platform; explore the best condition of detecting drugs/explosives hidden in human body through studying the X-ray diffraction recognition law by detecting drugs/explosives in the human environment on detection platform; provide theoretical foundation for studying new detecting principles and methods of using X-ray energy dispersion to detect drugs/explosives; provide reliable data for the data processing and recognition of energy dispersive X-ray detection.
(2) By using the energy dispersive X-ray spectrometer, this project studies the energy dispersive X-ray detection of drugs hidden in human body and its intelligent identification algorithm. Energy dispersive X-ray belongs to the weak X-ray beam, so energy dispersive X-ray detection's energy resolution is lower than conventional small angle scattering X-ray in similar X light source energy case. Besides, the detection target of this project is drugs/explosives hidden in human body. It is very difficult to identify materials through materials'diffraction peak when the scattered signal of the material under test is rather weak after background causing attenuation effect on the light beam.
Therefore, this project applies signal processing, pattern classification, statistical analysis and intelligent recognition and other algorithms to deal with the energy dispersive data obtained under complicated backgrounds. Using combinational algorithm of wavelet threshold and coherence function, we process the energy scattering X-ray data of heroin wrapped by different size and type of simulated human tissue (pork tissue). Through data denoising and coherence function matching, we establish simulated human tissue wrapped heroin's energy scattering X-ray detection database, and then identify it. In order to study the identification rate of different drugs and similar matter wrapped by simulated human tissue, this project chooses seven materials wrapped by simulated human tissue (pork tissue) including TNT, phenylacetic acid, methamphetamine, heroin, piperonal, human body and salt as test materials, uses PCA algorithm to identify them, optimizes PCA algorithm through the neural network and studies the influence law of the principal component number extracted from PCA algorithm on the recognition rate;
This project also chooses energy dispersive X-ray to detect drugs/explosives hidden in medical human body model, then adopts different pattern recognitions and statistical algorithms to process and recognize materials under test including TNT, phenylacetic acid, methamphetamine, heroin, piperonal, human body, and salt. The project puts forward a SVM classification algorithm based on particle swarm optimization. By the recognition and optimization of30sets of data of every material, we are seeking the optimal performance recognition algorithm of the SVM identification energy dispersive X-ray detecting drugs/explosives hidden in human body model. Furthermore, this project also studies a recognition algorithm of detecting drugs/explosives hidden in human body based on the discrete cosine transform and linear discriminant algorithm, and solves the problem of processing difficulty and low recognition of weak X-ray data by using this algorithm's advantage of extraction and classification of high dimensional data's decorrelation energy and feature.
At the end of this chapter, we studies the edge fisher analysis detecting drugs/explosives hidden in human body, especially without any demand on sample distribution, uses geometric relationships between the adjacent samples to represent the properties of the sample distribution, better solves the identification and classification problem of weak X-ray data, and applies the method of K-near neighbor (KNN), support vector machine (SVM), combined algorithm of PCA, PCA+LDA and MFA to raise the recognition effect of TNT, phenylacetic acid, methamphetamine, heroin, piperonal, human body, salt hidden in human body.
We research liquid dangerous materials'optical scattering characteristics under the detection of energy dispersive X-ray, choose the best test time, voltage, test angle, slit width etc. suitable for liquid dangerous materials, test common hazardous liquid sample, and get its3dimensional energy dispersive data. We build three-dimensional data model of liquid dangerous materials'energy-dispersive X-ray, regress, match and correct the load matrix's multivariate data in the three-dimensional model, establish three-dimensional PARAFAC data model of liquid dangerous materials' energy-dispersive X-ray, calculate the principal divisor number, then do the two-dimensional load matrix decomposition on the three-dimensional data model, classify and recognize the two-dimensional load matrix, and finally work out the accurate component and content of each material in the multi-component liquid explosive.
To sum up, in view of the energy dispersive X-ray spectrum characteristics of drugs/explosives in (human body or human body model) under the complex background condition, this project adopts the energy dispersive X-ray detection method to respectively studies the statistical model, pattern classification, recognition algorithm, matrix analysis and other function algorithm. On the basis of analysis and detection of common drugs/explosives and liquid dangerous materials, the project puts forward a variety of effective detection methods which may hopefully provide technical support for developing new pattern energy dispersive X-ray detection instruments.
引文
[1]李威.2012.基于化学计量学的危险有害物质光学检测技术的研究[D].合肥：中国科学技术大学.
[2]王同臻.2010.液体炸弹的危害性及其探测技术和仪器浅析[J].警用装备,5：66-69.
[3]张会中,伍继纯,王尧立.2008.铁路站车人体藏毒犯罪的特点及定性[J].铁道警官高等专科学校学报,18(2)：92-96.
[4]孟祥承.2003.爆炸物及毒品的探测技术[J].核电子学与探测技术,23(4)：371-377.
[5]李宁.2006.毒品和西药的THz光谱研究[D].北京：首都师范大学.
[6]Hiltmar Schubert, Andrey Kuznetsov.2007. Detection of Liquid Explosives and Flammable Agents in Connection with Terrorism [M]. Germany:Springer.
[7]A.V. Kuznetsov, O.I. Osetrov.2008. Overview of Liquid Explosives'Detection, Detection of Liquid EXplosives and Flammable Agents in Connection with Terrorism, NATO Science for Peace and Security Series B:Physics and Biophysics,138:7-13.
[8]阮书州.2011.高能X射线物质识别的蒙特卡罗模拟[D].长春：吉林大学.
[9]许峰,王海龙,关亚风.2005.离子迁移谱研究进展[J].化学进展,17(3)：53-62.
[10]J. R.Verkouteren, J. L.Staymates.2011. Reliability of ion mobility spectrometry for qualitative analysis of complex multicomponent illicit drug samples[J]. Forensic Science International,206 (1-3):190-196.
[11]S. P. Sharma, B. C.Purkait. S. C. Lahiri.2005. Qualitative and quantitative analysis of seized street drug samples and identification of source[J]. Forensic Science International,152(2-3): 235-240.
[12]T. Sasaki, Y. Makino.2006. Effective injection in pulsed splitles mode for impurity profiling of methamphetamine crystal by GC or GC/MS[J]. Forensic Science International,160 (1):1-10.
[13]黄魁,林远斌,吴腾芳,刘定.2007.国外爆炸物探测与识别技术综述[J].爆破器材,36(3)：34-38.
[14]许国旺等编著.2004.现代实用气相色谱法[M].化学工业出版社,北京.
[15]魏存峰,郁忠强,阎永廉等.2007.用于爆炸物探测的X射线相干散射技术研究[J].高能物理与核物理,31(1)：104-107.
[16]P. M.Joseph, R. D. Spital.1982. The effects of scatter in x-ray computed tomography [J]. Med. Phys.,9, (4),464-472.
[17]U. Neitzel, J. Kosanetzky, G Harding.1985. Coherent scatter in radiographic imaging:a Monte Carlo simulation study[J]. Phys. Med. Biol,30 (12):1289-1296.
[18]G Harding, J. Kosanetzky.1989. Scattered X-ray beam nondestructive testing[J]. Nuclear Instruments and Methods in Physics Research Section A,280 (2-3):517-528.
[19]G. Harding, M. Newton. J. Kosanetzky.1990. Energy-dispersive x-ray diffraction tomography[J]. Phys. Med. Biol.,35 (1):33-41.
[20]G Stix.1992. Explosive Images[J]. Scientific American,266(2):137-137.
[21]刘太辉,宋建中,刘景鑫等.2006.X射线相干散射成像在医学中的应用进展[J].北京生物医学工程,25(2)：204-208.
[22]仇满德.2004.薄膜外全反射角X射线能谱分析研究[D].石家庄：河北大学.
[23]钱丽英.2006. Al-Cu-Fe纳米准晶材料的制备和结构研究[D].南京：河海大学.
[24]杨在友.X射线束光纤线包特征参数测试系统设计与实现[D].南京：南京理工大学.
[25]李锋.2005. CdZnTe探测器在X射线荧光分析中的应用研究[D].长沙：国防科学技术大学.
[26]R. D.Luggar, M. J.Farquharson, et al.1998. Multivariate Analysis of Statistically Poor Spectra for the EDXRD Detection of Concealed Explosives[J]. X-ray Spectrometry,27 (2):87-94.
[27]B.Kampfe, F. Luczak, B.Michel.2005. Energy Dispersive X-Ray Diffraction[J]. Particle & Particle Systems Characterization.22 (6):391-396.
[28]张雨.2011.易制毒化学品X射线散射探测及其识别算法.合肥：中国科学技术大学.
[29]张芳.2013.基于EDXRS方法检测毒品和水中重金属离子的实验研究[D].合肥：中国科学技术大学.
[30]G Harding, J. Kosanetzky.1989. Scattered X-ray beam nondestructive testing[J]. Nuclear Instruments and Methods in Physics Research Section A,280(2-3):517-528.
[31]S. Pani, E. Cook, et al.2009. Modelling an Energy-Dispersive X-Ray Diffraction System for Drug Detection[J]. IEEE Trans. Nucl. Sci.,56 (3):1238-1241.
[32]Harding, G, Optimization criteria for CXRS baggage inspection. Proc. SPIE,1995,2511, 64-70.
[33]Harding, G, X-ray diffraction imaging-A multi-generational perspective. Appl. Radiat. Isot., 2009,67, (2),287-295.
[34]Harding, G.; Schreiber, B., Coherent X-ray scatter imaging and its applications in biomedical science and industry. Radiation Physics and Chemistry,1999,56, (1-2),229-245.
[35]江超华.1980.多晶X光衍射仪实验技术[J].分析仪器,1：15-14.
[36]李建欣.2009.XRD全谱拟合精修对贵州煤中矿物质的定量研究[D].焦作：河南理工大学.
[37]T. Eggert, O. Boslau, et al.2004. Goldstrass, P.; Kemmer, J., Silicon drift detectors with enlarged sensitive areas[J]. X-Ray Spectrometry,33 (4):246-252.
[38]M. Ferretti.2004. Fluorescence from the collimator in Si-PIN and Si-drift detectors:problems and solutions for the XRF analysis of archaeological and historical materials[J]. Nuclear Instruments and Methods in Physics Research Section B,226 (3):453-460.
[39]P. Lechner, A.Pahlke, H.Soltau.2004. Novel high-resolution silicon drift detectors[D]. X-Ray Spectrometry,33(4):256-261.
[40]罗立强.2006.非常规X射线能量探测技术[J].岩矿测试,25(1)：49-54.
[41]朱传新等.2006.高纯锗探测器探测效率研究[J].核电子学与探测技术,26(2)：191-194.
[42]C. H. Malden, D.Speller.2000. A CdZnTE array for the detection of explosives in baggage by energy-dispersive X-ray diffraction signatures at multiple scatter angles[J]. Nuclear Instruments and Methods in Physics Research A,449(1-2):408-415.
[43]E. M.Pashaev, V. N.Peregudov, et al.2003. The effect of X-ray radiation on the structural and electrical properties of CdZnTe solid solution[J]. Physica Status Solidi (C),0(3):897-901.
[44]王鹏.2003.能量色散X荧光分析仪的研制[D].成都：四川大学硕士论文.
[45]吴春蓉.2007.便携式X射线荧光分析仪的放大电路研制.成都：成都理工大学.
[46]孙柏等.2009.能量色散X射线散射实验平台研制及其在违禁品检测中的应用研究[D].核技术,32(11)：859-863.
[47]李志刚,汤博.2013.基于MATLAB下BP神经网络的边坡稳定性预测[J].水科学与工程技术,5：25-30.
[48]李怀珍,邓广涛.2006.岩质边坡稳定性预测的BP网络法[J].地质灾害与环境保护,3：106-109.
[49]胡求索.2007.基于人工神经网络的软件质量预测模型研究[D].桂林：广西大学.
[50]李进如.2008.基于改进小波神经网络的故障识别方法研究[D].北京：华北电力大学.
[51]宗岳.2009.某火炮电液伺服系统的新型控制策略研究[D].南京：南京理工大学.
[52]颜浩.2010.精炼炉硫含量预报模型的研究.沈阳：东北大学.
[53]任若恩,王惠文.1997.多元统计数据分析-理论、方法、实例[M].北京：国防工业出版社.
[54]张浩然,韩正之,李昌刚.2002.支持向量机[J].计算机科学,29(12)：135-142.
[55]Burges C.J.C.1998. A tutorial on support vector machines for pattern recognition.Data Min. Knowl. Disc,2:121-167.
[56]Hsu CW, Lin, CJ.2002. A comparison of methods for multiclass support vector machines. IEEE T rans.Neural.Networ,13(2),415-425.
[57]潘明清等.基于主元分析的支持向量数据描述机械故障诊断[J].传感技术学报,19(1)：128-131.
[58]Chen PW, Wang JY, Lee HM. Model selection of SVMs using GA approach[C]//Proc of 2004 IEEE Int Joint Conf on Neural Networks. Piscataway, NJ:IEEE Press,2004:2035-2040.
[59]Zheng Chunhong, Jiao Licheng. Automatic parameters selection for SVM based on GA[C]// Proc of the 5th World Congress on Intelligent Control and Automation. Piscataway, NJ:IEEE Press,2004:1869-1872.
[60]Osvaldo A. Rosso, Susana Blanco, et al.2001. Wavelet entropy:a new tool for analysis of short duration brain electrical signals[J]. J. Neurosci.Methods,105 (21) 65-75.
[61]邵信光,杨慧中,陈刚.2006.基于粒子群优化算法的支持向量机参数选择及其应用[J].控制理论与应用,23(5)：740-748.
[62]E. Mayoraz, E. Alpaydin.1999. Support vector machines for multi-class classification[J]. Eng. Appl. Bio-Inspired Artif. Neural Networks,21(2) 833-842.
[63]Rocha A, Goldenstein S.2009. Multi-class from binary:drivide to conquer. VISAPP,1-8.
[64]Hsu CW, Lin, CJ.2002. A comparison of methods for multiclass support vector machines. IEEE T rans.Neural.Networ,13(2),415-425.
[65]蔡敏芹等.2005.中国人辐照仿真人体模型及人体器官数学模型.中国医学影像技术.21(9)：1464-1467.
[66]林大全,吴大可.1998.仿真辐照人体模型的现状与应用[J].世界医疗器械.4(8)：38-42.
[67]N. Ahmed, T. Natarajan and K. R. Rao.1974. Discrete cosine transform[j]. IEEE Trans. Comput.1974,23(1):90-93.
[68]崔伟东,周志华,李星.2001.支持向量机研究[J].计算机工程与应用,1：58-61.
[69]史峰,王小川,郁磊,李洋2010. Matlab神经网络30个案例分析.北京：北京航空航天大学出版社.
[70]张瑞强.2012.基于离散余弦变换的时间序列分类问题研究[D].杭州：浙江工业大学.
[71]黄宇.2006.车辆目标类型识别算法研究与实现[D].成都：电子科技大学.
[72]姜园媛.2012.人脸识别中关于线性判别分析算法的若干研究[D].吉林：吉林大学.
[73]胡海峰等.2009.液体安全探测技术综述[J].中国体视学与图像分析,3：321-325.
[74]Leon Barron, Elizabeth Gilchrist, Ion chromatography-mass spectrometry:A review of recent technologies and applications in forensic and environmental eXplosives analysis, Analytica Chimica Acta, Vol.806 (2), pp.27-54, Jan.2014
[75]Itamar Malka, Alona Petrushansky, Salman Rosenwaks, Ilana Bar, Detection of Explosives and latent fingerprint residues utilizing laser pointer-based Raman spectroscopy, Appl. Phys. B, Vol.113, pp.511-518,2013
[76]Gradisek, Anton, Apih, Tomaz, NMR-Based Liquid EXplosives Detector, Applied Magnetic Resonance, Vol.38(4),485-493,2010
[77]Eliasson C., Macleod N. A., Matousek P., Noninvasive detection of concealed liquid eXplosives using Raman spectroscopy, Analytical Chemistry, Vol.79, pp.8185-8189,2007
[78]陈力,炸药探测技术分类与发展现状分析,中国安防,第7期,22-27页,2008
[79]成云飞,吴轶轩,谢裕康,便携式离子迁移谱炸药探测仪,警察技术,第6期,62-64页,2008
[80]胡颖,张存林,郭澜涛,王晓红,THz光谱技术在炸药及其相关材料探测中的应用进展,物理,第36卷,第1期,68-72页,2007
[81]蔡建刚,液体炸药的特性及其探测技术,武警学院学报,第1期,25-28页,2009
[82]清华大学,清华同方威视技术股份有限公司,一种用放射源对液体进行背散射安全检测的方法及其装置,2006.5,中国发明专利,CN200410009895.1.
[83]清华大学,清华同方威视技术股份有限公司,一种用放射源对液体进行安全检测的方法及其装置,2006.5,中国发明专利,CN200410009896.6.
[84]清华大学,清华同方威视技术股份有限公司,一种用射线源对液体进行CT安全探测的方法及其装置,2006.5,中国发明专利CN200410009897.0.
[85]清华大学,清华同方威视技术股份有限公司,一种利用射线源对液体物品进行CT安检查的设备,2007.3,中国发明专利,CN200510086458.4.
[86]公安部第一研究所,北京中盾安民分析技术有限公司,液体安全探测的双能量X射线螺旋CT系统,2009.7,中国发明专利,200810116659.8.
[87]公安部第一研究所,北京中盾安民分析技术有限公司,液体安全探测的双能量X射线螺 旋CT装置及其探测方法,2010.1,中国发明专利,
[88]A.V. Kuznetsov and O.I. Osetrov, Overview of Liquid EXplosives'Detection, Detection of Liquid EXplosives and Flammable Agents in Connection with Terrorism, NATO Science for Peace and Security Series B:Physics and Biophysics, pp.7-13,2008
[89]A. V. Evsenin, A.V. Kuznetsov, O. I. Osetrov, D. N. Vakhtin, Detection of Hidden EXplosives by Nanosecond Neutron Analysis Technique, Detection of Bulk EXplosives Advanced Techniques against Terrorism, NATO Science Series, Vol.138, pp.89-103,2004,
[90]Yu Zhong, Minqiang Li*. Bai Sun, Jin Wang, Fang Zhang, Daovang Yu. Yu Zhang, Jinhuai Liu*, Non-invasive investigation of liquid materials using energy dispersive X-ray scattering, Measurement, Vol.45(6), pp.1540-1546,2012
[91]Yu Zhong, Fang Zhang, Wei Li, Minaiang Li. Bai Sun, Yu Zhang, Daovang Yu*, Jinhuai Liu*, A new application of X-ray scattering using principal component analysis-classification and identification of liquid precursor chemicals, X-Ray Spectrometry, Vol.42(1), pp.45-51, 2013
[92]Bro R., PARAFAC. Tutorial and applications, Chemometrics and Intelligent Laboratory Systems, Vol.38(2), pp.149-171,1997
[93]Hailong Wu, Jinfang Nie, Yongjie Yu, Rulin Yu, Multi-way chemometric methodologies and applications:A central summary of our research work, Analytica Chimica Acta, Vol.650(1), pp.131-142, (2009)
[94]T. Do, N. S. McIntyre, R. A. Harshman, M. E. Lundy, S. J. Splinter, Application of Parallel Factor Analysis and X-ray Photoelectron Spectroscopy to the Initial Stages in OXidation of Aluminium I. The A1 2p Photoelectron Line, Surf. Interface Anal. Vol.27(12), pp.618-628, 1999
[95]R.Bro.1997. PARAFAC. Tutorial and applications[J]. Chemometrics and Intelligent
Laboratory Systems,38(2):149-171.
[96]R.A. Harshman.1970.Foundation of the PARAFAC procedure:Models and condition for an 'explanatory'multi-model factor analysis[J].UCLA Working paper in Phonetics,16(1):1-84.
[97]韩清娟.2008.多维光谱数据解析的化学计量学算法及应用研究[D].长沙：湖南大学.
[98]朱绍华.2008..高维化学计量学方法的研究及其在农药分析方面的应用[D].长沙：湖南大学.
[2]王同臻.2010.液体炸弹的危害性及其探测技术和仪器浅析[J].警用装备,5：66-69.
[3]张会中,伍继纯,王尧立.2008.铁路站车人体藏毒犯罪的特点及定性[J].铁道警官高等专科学校学报,18(2)：92-96.
[4]孟祥承.2003.爆炸物及毒品的探测技术[J].核电子学与探测技术,23(4)：371-377.
[5]李宁.2006.毒品和西药的THz光谱研究[D].北京：首都师范大学.
[6]Hiltmar Schubert, Andrey Kuznetsov.2007. Detection of Liquid Explosives and Flammable Agents in Connection with Terrorism [M]. Germany:Springer.
[7]A.V. Kuznetsov, O.I. Osetrov.2008. Overview of Liquid Explosives'Detection, Detection of Liquid EXplosives and Flammable Agents in Connection with Terrorism, NATO Science for Peace and Security Series B:Physics and Biophysics,138:7-13.
[8]阮书州.2011.高能X射线物质识别的蒙特卡罗模拟[D].长春：吉林大学.
[9]许峰,王海龙,关亚风.2005.离子迁移谱研究进展[J].化学进展,17(3)：53-62.
[10]J. R.Verkouteren, J. L.Staymates.2011. Reliability of ion mobility spectrometry for qualitative analysis of complex multicomponent illicit drug samples[J]. Forensic Science International,206 (1-3):190-196.
[11]S. P. Sharma, B. C.Purkait. S. C. Lahiri.2005. Qualitative and quantitative analysis of seized street drug samples and identification of source[J]. Forensic Science International,152(2-3): 235-240.
[12]T. Sasaki, Y. Makino.2006. Effective injection in pulsed splitles mode for impurity profiling of methamphetamine crystal by GC or GC/MS[J]. Forensic Science International,160 (1):1-10.
[13]黄魁,林远斌,吴腾芳,刘定.2007.国外爆炸物探测与识别技术综述[J].爆破器材,36(3)：34-38.
[14]许国旺等编著.2004.现代实用气相色谱法[M].化学工业出版社,北京.
[15]魏存峰,郁忠强,阎永廉等.2007.用于爆炸物探测的X射线相干散射技术研究[J].高能物理与核物理,31(1)：104-107.
[16]P. M.Joseph, R. D. Spital.1982. The effects of scatter in x-ray computed tomography [J]. Med. Phys.,9, (4),464-472.
[17]U. Neitzel, J. Kosanetzky, G Harding.1985. Coherent scatter in radiographic imaging:a Monte Carlo simulation study[J]. Phys. Med. Biol,30 (12):1289-1296.
[18]G Harding, J. Kosanetzky.1989. Scattered X-ray beam nondestructive testing[J]. Nuclear Instruments and Methods in Physics Research Section A,280 (2-3):517-528.
[19]G. Harding, M. Newton. J. Kosanetzky.1990. Energy-dispersive x-ray diffraction tomography[J]. Phys. Med. Biol.,35 (1):33-41.
[20]G Stix.1992. Explosive Images[J]. Scientific American,266(2):137-137.
[21]刘太辉,宋建中,刘景鑫等.2006.X射线相干散射成像在医学中的应用进展[J].北京生物医学工程,25(2)：204-208.
[22]仇满德.2004.薄膜外全反射角X射线能谱分析研究[D].石家庄：河北大学.
[23]钱丽英.2006. Al-Cu-Fe纳米准晶材料的制备和结构研究[D].南京：河海大学.
[24]杨在友.X射线束光纤线包特征参数测试系统设计与实现[D].南京：南京理工大学.
[25]李锋.2005. CdZnTe探测器在X射线荧光分析中的应用研究[D].长沙：国防科学技术大学.
[26]R. D.Luggar, M. J.Farquharson, et al.1998. Multivariate Analysis of Statistically Poor Spectra for the EDXRD Detection of Concealed Explosives[J]. X-ray Spectrometry,27 (2):87-94.
[27]B.Kampfe, F. Luczak, B.Michel.2005. Energy Dispersive X-Ray Diffraction[J]. Particle & Particle Systems Characterization.22 (6):391-396.
[28]张雨.2011.易制毒化学品X射线散射探测及其识别算法.合肥：中国科学技术大学.
[29]张芳.2013.基于EDXRS方法检测毒品和水中重金属离子的实验研究[D].合肥：中国科学技术大学.
[30]G Harding, J. Kosanetzky.1989. Scattered X-ray beam nondestructive testing[J]. Nuclear Instruments and Methods in Physics Research Section A,280(2-3):517-528.
[31]S. Pani, E. Cook, et al.2009. Modelling an Energy-Dispersive X-Ray Diffraction System for Drug Detection[J]. IEEE Trans. Nucl. Sci.,56 (3):1238-1241.
[32]Harding, G, Optimization criteria for CXRS baggage inspection. Proc. SPIE,1995,2511, 64-70.
[33]Harding, G, X-ray diffraction imaging-A multi-generational perspective. Appl. Radiat. Isot., 2009,67, (2),287-295.
[34]Harding, G.; Schreiber, B., Coherent X-ray scatter imaging and its applications in biomedical science and industry. Radiation Physics and Chemistry,1999,56, (1-2),229-245.
[35]江超华.1980.多晶X光衍射仪实验技术[J].分析仪器,1：15-14.
[36]李建欣.2009.XRD全谱拟合精修对贵州煤中矿物质的定量研究[D].焦作：河南理工大学.
[37]T. Eggert, O. Boslau, et al.2004. Goldstrass, P.; Kemmer, J., Silicon drift detectors with enlarged sensitive areas[J]. X-Ray Spectrometry,33 (4):246-252.
[38]M. Ferretti.2004. Fluorescence from the collimator in Si-PIN and Si-drift detectors:problems and solutions for the XRF analysis of archaeological and historical materials[J]. Nuclear Instruments and Methods in Physics Research Section B,226 (3):453-460.
[39]P. Lechner, A.Pahlke, H.Soltau.2004. Novel high-resolution silicon drift detectors[D]. X-Ray Spectrometry,33(4):256-261.
[40]罗立强.2006.非常规X射线能量探测技术[J].岩矿测试,25(1)：49-54.
[41]朱传新等.2006.高纯锗探测器探测效率研究[J].核电子学与探测技术,26(2)：191-194.
[42]C. H. Malden, D.Speller.2000. A CdZnTE array for the detection of explosives in baggage by energy-dispersive X-ray diffraction signatures at multiple scatter angles[J]. Nuclear Instruments and Methods in Physics Research A,449(1-2):408-415.
[43]E. M.Pashaev, V. N.Peregudov, et al.2003. The effect of X-ray radiation on the structural and electrical properties of CdZnTe solid solution[J]. Physica Status Solidi (C),0(3):897-901.
[44]王鹏.2003.能量色散X荧光分析仪的研制[D].成都：四川大学硕士论文.
[45]吴春蓉.2007.便携式X射线荧光分析仪的放大电路研制.成都：成都理工大学.
[46]孙柏等.2009.能量色散X射线散射实验平台研制及其在违禁品检测中的应用研究[D].核技术,32(11)：859-863.
[47]李志刚,汤博.2013.基于MATLAB下BP神经网络的边坡稳定性预测[J].水科学与工程技术,5：25-30.
[48]李怀珍,邓广涛.2006.岩质边坡稳定性预测的BP网络法[J].地质灾害与环境保护,3：106-109.
[49]胡求索.2007.基于人工神经网络的软件质量预测模型研究[D].桂林：广西大学.
[50]李进如.2008.基于改进小波神经网络的故障识别方法研究[D].北京：华北电力大学.
[51]宗岳.2009.某火炮电液伺服系统的新型控制策略研究[D].南京：南京理工大学.
[52]颜浩.2010.精炼炉硫含量预报模型的研究.沈阳：东北大学.
[53]任若恩,王惠文.1997.多元统计数据分析-理论、方法、实例[M].北京：国防工业出版社.
[54]张浩然,韩正之,李昌刚.2002.支持向量机[J].计算机科学,29(12)：135-142.
[55]Burges C.J.C.1998. A tutorial on support vector machines for pattern recognition.Data Min. Knowl. Disc,2:121-167.
[56]Hsu CW, Lin, CJ.2002. A comparison of methods for multiclass support vector machines. IEEE T rans.Neural.Networ,13(2),415-425.
[57]潘明清等.基于主元分析的支持向量数据描述机械故障诊断[J].传感技术学报,19(1)：128-131.
[58]Chen PW, Wang JY, Lee HM. Model selection of SVMs using GA approach[C]//Proc of 2004 IEEE Int Joint Conf on Neural Networks. Piscataway, NJ:IEEE Press,2004:2035-2040.
[59]Zheng Chunhong, Jiao Licheng. Automatic parameters selection for SVM based on GA[C]// Proc of the 5th World Congress on Intelligent Control and Automation. Piscataway, NJ:IEEE Press,2004:1869-1872.
[60]Osvaldo A. Rosso, Susana Blanco, et al.2001. Wavelet entropy:a new tool for analysis of short duration brain electrical signals[J]. J. Neurosci.Methods,105 (21) 65-75.
[61]邵信光,杨慧中,陈刚.2006.基于粒子群优化算法的支持向量机参数选择及其应用[J].控制理论与应用,23(5)：740-748.
[62]E. Mayoraz, E. Alpaydin.1999. Support vector machines for multi-class classification[J]. Eng. Appl. Bio-Inspired Artif. Neural Networks,21(2) 833-842.
[63]Rocha A, Goldenstein S.2009. Multi-class from binary:drivide to conquer. VISAPP,1-8.
[64]Hsu CW, Lin, CJ.2002. A comparison of methods for multiclass support vector machines. IEEE T rans.Neural.Networ,13(2),415-425.
[65]蔡敏芹等.2005.中国人辐照仿真人体模型及人体器官数学模型.中国医学影像技术.21(9)：1464-1467.
[66]林大全,吴大可.1998.仿真辐照人体模型的现状与应用[J].世界医疗器械.4(8)：38-42.
[67]N. Ahmed, T. Natarajan and K. R. Rao.1974. Discrete cosine transform[j]. IEEE Trans. Comput.1974,23(1):90-93.
[68]崔伟东,周志华,李星.2001.支持向量机研究[J].计算机工程与应用,1：58-61.
[69]史峰,王小川,郁磊,李洋2010. Matlab神经网络30个案例分析.北京：北京航空航天大学出版社.
[70]张瑞强.2012.基于离散余弦变换的时间序列分类问题研究[D].杭州：浙江工业大学.
[71]黄宇.2006.车辆目标类型识别算法研究与实现[D].成都：电子科技大学.
[72]姜园媛.2012.人脸识别中关于线性判别分析算法的若干研究[D].吉林：吉林大学.
[73]胡海峰等.2009.液体安全探测技术综述[J].中国体视学与图像分析,3：321-325.
[74]Leon Barron, Elizabeth Gilchrist, Ion chromatography-mass spectrometry:A review of recent technologies and applications in forensic and environmental eXplosives analysis, Analytica Chimica Acta, Vol.806 (2), pp.27-54, Jan.2014
[75]Itamar Malka, Alona Petrushansky, Salman Rosenwaks, Ilana Bar, Detection of Explosives and latent fingerprint residues utilizing laser pointer-based Raman spectroscopy, Appl. Phys. B, Vol.113, pp.511-518,2013
[76]Gradisek, Anton, Apih, Tomaz, NMR-Based Liquid EXplosives Detector, Applied Magnetic Resonance, Vol.38(4),485-493,2010
[77]Eliasson C., Macleod N. A., Matousek P., Noninvasive detection of concealed liquid eXplosives using Raman spectroscopy, Analytical Chemistry, Vol.79, pp.8185-8189,2007
[78]陈力,炸药探测技术分类与发展现状分析,中国安防,第7期,22-27页,2008
[79]成云飞,吴轶轩,谢裕康,便携式离子迁移谱炸药探测仪,警察技术,第6期,62-64页,2008
[80]胡颖,张存林,郭澜涛,王晓红,THz光谱技术在炸药及其相关材料探测中的应用进展,物理,第36卷,第1期,68-72页,2007
[81]蔡建刚,液体炸药的特性及其探测技术,武警学院学报,第1期,25-28页,2009
[82]清华大学,清华同方威视技术股份有限公司,一种用放射源对液体进行背散射安全检测的方法及其装置,2006.5,中国发明专利,CN200410009895.1.
[83]清华大学,清华同方威视技术股份有限公司,一种用放射源对液体进行安全检测的方法及其装置,2006.5,中国发明专利,CN200410009896.6.
[84]清华大学,清华同方威视技术股份有限公司,一种用射线源对液体进行CT安全探测的方法及其装置,2006.5,中国发明专利CN200410009897.0.
[85]清华大学,清华同方威视技术股份有限公司,一种利用射线源对液体物品进行CT安检查的设备,2007.3,中国发明专利,CN200510086458.4.
[86]公安部第一研究所,北京中盾安民分析技术有限公司,液体安全探测的双能量X射线螺旋CT系统,2009.7,中国发明专利,200810116659.8.
[87]公安部第一研究所,北京中盾安民分析技术有限公司,液体安全探测的双能量X射线螺 旋CT装置及其探测方法,2010.1,中国发明专利,
[88]A.V. Kuznetsov and O.I. Osetrov, Overview of Liquid EXplosives'Detection, Detection of Liquid EXplosives and Flammable Agents in Connection with Terrorism, NATO Science for Peace and Security Series B:Physics and Biophysics, pp.7-13,2008
[89]A. V. Evsenin, A.V. Kuznetsov, O. I. Osetrov, D. N. Vakhtin, Detection of Hidden EXplosives by Nanosecond Neutron Analysis Technique, Detection of Bulk EXplosives Advanced Techniques against Terrorism, NATO Science Series, Vol.138, pp.89-103,2004,
[90]Yu Zhong, Minqiang Li*. Bai Sun, Jin Wang, Fang Zhang, Daovang Yu. Yu Zhang, Jinhuai Liu*, Non-invasive investigation of liquid materials using energy dispersive X-ray scattering, Measurement, Vol.45(6), pp.1540-1546,2012
[91]Yu Zhong, Fang Zhang, Wei Li, Minaiang Li. Bai Sun, Yu Zhang, Daovang Yu*, Jinhuai Liu*, A new application of X-ray scattering using principal component analysis-classification and identification of liquid precursor chemicals, X-Ray Spectrometry, Vol.42(1), pp.45-51, 2013
[92]Bro R., PARAFAC. Tutorial and applications, Chemometrics and Intelligent Laboratory Systems, Vol.38(2), pp.149-171,1997
[93]Hailong Wu, Jinfang Nie, Yongjie Yu, Rulin Yu, Multi-way chemometric methodologies and applications:A central summary of our research work, Analytica Chimica Acta, Vol.650(1), pp.131-142, (2009)
[94]T. Do, N. S. McIntyre, R. A. Harshman, M. E. Lundy, S. J. Splinter, Application of Parallel Factor Analysis and X-ray Photoelectron Spectroscopy to the Initial Stages in OXidation of Aluminium I. The A1 2p Photoelectron Line, Surf. Interface Anal. Vol.27(12), pp.618-628, 1999
[95]R.Bro.1997. PARAFAC. Tutorial and applications[J]. Chemometrics and Intelligent
Laboratory Systems,38(2):149-171.
[96]R.A. Harshman.1970.Foundation of the PARAFAC procedure:Models and condition for an 'explanatory'multi-model factor analysis[J].UCLA Working paper in Phonetics,16(1):1-84.
[97]韩清娟.2008.多维光谱数据解析的化学计量学算法及应用研究[D].长沙：湖南大学.
[98]朱绍华.2008..高维化学计量学方法的研究及其在农药分析方面的应用[D].长沙：湖南大学.