Research On Identification Of Hazardous Substance Based On Terahertz Spectrums

In air transportation,train freight,customs inspection,etc.,security inspection based on hazardous goods,contraband and other substances is an indispensable portion.At present,the safety inspection in these places is mainly carried out based on X-ray imaging to discriminate the image manually.Since X-ray has certain harm to the human body,to seek a detection and identification method with higher security and better recognition effect has become one of the hot issues that many researchers are paying attention to.While compared with X-rays,the energy of terahertz waves is lower,so it is safer for humans.And as a highly penetrating radiation,terahertz waves can provide the possibility to identify packaged materials.Aiming at the detection and identification of hazardous organic compound,an identification method based on terahertz spectrum is proposed.The main works are as follows:(1)For terahertz time-domain spectral data of organic compounds,an identification method based on differential-PCA(Principal Component Analysis)-SVM(Support Vector Machine)is proposed.Firstly,the absorption coefficient spectrum is calculated according to the terahertz time domain signal of the sample.Then,the data features in the frequency range of 0.2-2.5 THz is extracted.During the feature extraction,an expansion of sample data amount method based on differential data is proposed,and is combined with the principal component analysis method to achieve feature extraction.Finally,the support vector machine is used to establish the mathematical model,and the unknown samples are identified with that.In the experiment,the terahertz spectral data of 15 organic compounds were identified with the method,and the correct recognition rate was 93.33%.(2)For the identification of hazardous organic compounds concealed by materials,a template matching method based on spectral angle is proposed.Firstly,the data is divided into two groups according to whether the sample is concealed by the material.Unconcealed pure material data are viewed as a standard database,while the feature vector of it is extracted and is regarded as the standard template vector.The concealed substance is to be identified.The feature vector of it is extracted with the same method,and is compared with the vectors in the standard database to determine the category.The experimental results show that the spectral angle matching method based on absorption coefficient spectrum performed better in recognition,which can reduce the influence of the recognition effect which is brought by amplitude decline that is caused by materials’ absorption of terahertz waves.