危险品检测算法的研究与实现

发布时间：2018-04-01 19:33

本文选题：危险品检测　切入点：FastICA算法　出处：《成都理工大学》2016年硕士论文

【摘要】：近年来,国际形势动荡,恐怖袭击事件时有发生。公共安全问题日趋严重,对于公交车上发生的危害公共安全的事件也越来越多,犯罪分子常常采用汽油、管制刀具等对普通民众进行伤害,造成的社会影响极其恶劣。因此为了在检测到的复杂混合信号中,将危险品信号分离出来,研究一种有效的盲源分离算法是非常重要的。盲源分离研究的是在不知道源信号和将其混合的方式情况下,仅通过源信号的一些统计特性,然后根据观测信号(混合信号)来恢复出源信号。独立分量分析是盲源分离方法中的一种特殊算法,它是建立在源信号相互独立情况下的一种算法。而快速独立分量分析(FastICA)算法是利用了最大化信号的非高斯性,该算法采用了牛顿方法去处理计算值的多个采样数据,每步的迭代计算能从混合信号中间分离出一个独立的成分,属于独立分量分析算法中的一种快速计算方法。在本论文中,首先分析了危险品微波检测的原理,根据检测的原理对检测系统的功能进行了实现。其中以汽油、乙醇等常见危险品的介电常数为基础,远距离非接触式测量危险品S参数,并对S参数进行分析,实现利用厘米波远距离非接触式地排查危险品。然后对算法进行了研究,针对不同的情况使用了两种算:累加算法和Fast ICA算法。在只检测到一种危险品的时候,我们只需要判断出它是什么危险品,论文中分别介绍了累积求和与累加算法来进行判断,最终根据得出的结果判断出累加算法比累积求和的算法更加适用于危险品检测中;对于同时检测到多种危险品信号的时候,由于此时的信号是一个观测信号,因而我们需要利用Fast ICA算法对其进行盲源分离,得到不同危险品的独立信号。在论文中分别以水和空气作为参考,通过对比发现以水为参考得出的解混信号与源信号的相似程度要比空气的要加接近。为了将危险品检测的算法进行实现,设计了一个危险品检测软件系统,系统包括上位机部分与下位机部分。开发危险品检测软件系统时,对现有的一些可以开发界面的工具进行了一个简单的对比,最终选择了Qt开发框架。在通信方面,采用高可靠性TCP协议的客户端/服务器模式来进行通信,在下位机请求连接上位机并连接成功时,就可以开始下位机信息的发送和上位机数据的接收。当上位机接收检测到危险品的信息后,就会在界面上显示危险品图片和对图片的说明,发出报警声,并且还能将接收到的信息进行截图保存。在危险品检测算法的研究与实现的过程中,对算法和程序进行了不断的改进和调试,改正了各种细节错误。初步实现了能够将检测到的危险品进行判别,并通过危险品检测软件系统来实现报警,在通信的稳定性以及上位机和下位机各自的功能等方面都基本达到了要求。
[Abstract]:In recent years, the international situation has been turbulent, terrorist attacks have occurred from time to time. Public safety problems are becoming more and more serious, and there are more and more incidents that endanger public safety on buses. Criminals often use gasoline. In order to separate dangerous signals from the complex mixed signals detected, the social impact caused by the harm done to ordinary people by controlling knives and others is extremely bad. It is very important to study an effective blind source separation algorithm. Blind source separation studies only some statistical characteristics of the source signal without knowing the source signal and mixing it. Independent component analysis (ICA) is a special algorithm in blind source separation. The fast independent component analysis (FastICA) algorithm takes advantage of the non-#china_person0# property of the maximized signal, and the Newton method is used to deal with multiple sampled data of the calculated value, the algorithm is based on the source signal being independent of each other, and the fast independent component analysis (FastICA) algorithm takes advantage of the non-#china_person0# property of the maximized signal. The iterative calculation of each step can separate an independent component from the mixed signal, which is a fast calculation method in the independent component analysis algorithm. In this paper, the principle of microwave detection of dangerous goods is first analyzed. According to the principle of detection, the function of the detection system is realized. Based on the dielectric constant of gasoline, ethanol and other common dangerous goods, the S parameters of dangerous goods are measured and analyzed by remote non-contact method. In this paper, centimeter wave is used to detect dangerous goods in a long distance and non-contact way. Then, the algorithm is studied, and two kinds of algorithms are used for different cases: accumulative algorithm and Fast ICA algorithm. When only one kind of dangerous goods is detected, when only one kind of dangerous goods is detected, We only need to judge what dangerous goods it is. In this paper, the cumulative summation and accumulation algorithms are introduced to judge. Finally, it is concluded that the accumulative algorithm is more suitable for dangerous goods detection than the cumulative summation algorithm. When a variety of dangerous goods signals are detected at the same time, because the signal is an observational signal, we need to use the Fast ICA algorithm to separate its blind sources. Get the independent signal of different dangerous goods. In the paper, water and air are respectively used as reference, It is found by comparison that the similarity between the decontamination signal and the source signal obtained by reference to water is closer than that of air. In order to realize the algorithm of dangerous goods detection, a dangerous goods detection software system is designed. The system includes the upper computer part and the lower computer part. When developing the dangerous goods detection software system, the paper makes a simple comparison to some existing tools that can develop interface, and finally chooses the QT development framework. The client / server mode of high reliability TCP protocol is used to communicate. When the lower computer requests to connect with the host computer and the connection is successful, When the host computer receives the information of the dangerous goods, it will display the picture of dangerous goods and explain the picture on the interface, and issue the alarm sound. In the process of research and implementation of dangerous goods detection algorithm, the algorithm and program have been continuously improved and debugged. Has corrected all kinds of details error. Has realized preliminarily can detect the dangerous goods to carry on the discrimination, and through the dangerous goods detection software system realizes the alarm, The stability of communication and the functions of upper computer and lower computer have basically met the requirements.
【学位授予单位】：成都理工大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：D63;TN911.7

【相似文献】