Semi-automated technique for trace detection of explosives using hyperspectral sensor | |
| Author | Chaudhary, Siddharth |
| Call Number | AIT Diss. no.RS-20-03 |
| Subject(s) | Hyperspectral imaging--Data processing Remote sensing--Data processing Explosives--Detection |
| Note | A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Remote Sensing and Geographic Information Systems, School of Engineering and Technology |
| Publisher | Asian Institute of Technology |
| Abstract | The threat of explosive is increasing due to the large number of conflicts taking place across the world due to which there is an increase in number of warzones. This is a serious issue that is affecting socio-econorny of many countries like public security, unused arable land, closing of trade routes, isolation of villages. Such problems act as a hindrance in the development of the country. Frequency of the terrorist activities have increased in the last two decades and leading to a global threat which is challenging the humanity. As reported by AOA V "In 2017 42,972 deaths and injuries were recorded from the use of explosive weapons around the world and of those harmed, 74% (31904) were reported to be civilians". In 2017 death of civilians was highest with an increase of38% from 2016 and 165% from 2011. Explosive violence affects dozens of countries around the world. Most of the terrorist attacks use a special type of bomb known as Improvised Explosive Devices (lED) in which the explosives are stored inside metal containers. The explosives are made up of chemical compounds which can have high impact even if used in less quantities. lED's can be grouped as military, commercial and homemade based on materials used for manufacturing them. These problems motivate the government and research community to develop a technique for fast and accurate detection of explosives. Though there have been number of researches focused on trace detection of chemicals used in explosives where researchers have come up with new techniques like Raman spectroscopy, laser induced breakdown spectroscopy, laser induced fluorescence and ion mobility spectroscopy. But very few works have been done in standoff trace detection using hyperspectral imaging system. The current research techniques used for standoff trace detection involves several difficulties and challenges due to physical properties of the traces. The aim of this study was to develop a spectral library for different chemicals used in explosives based on their spectral reflectance response, investigate the potential of non-destructive hyperspectral imaging system (1) and accuracy of the model developed using Support Vector Machine (SVM) and Artificial Neural Network (ANN) for determining trace detection of explosives. Raman spectroscopy has been used in similar studies, but no study has been published which is based on measurement of reflectance from hyperspectral sensor for trace detection of explosives. I used in this study has an advantage over existing techniques due to its combination of imaging system and spectroscopy, along with being contactless and non-destructive in nature. Hyperspectral images of the chemical were collected using BaySpec hyperspectral sensor which operated in the spectral range of 400 - 1000 nm (144 bands). Image processing was applied on the acquired hyperspectral image to select the region of interest (ROI) and to extract the spectral reflectance of the chemicals which were stored as spectral library. Principal Component Analysis (PCA) and first derivative were applied to reduce the high dimensionality of the image and to determine the optimal wavelengths 'between 400 - 1000 nm. In total, 22 out of 144 wavelengths were selected by analysing the loadings of principal components (PC). SVM was used to develop the classification model. SVM model established on the whole spectrum from 400 - 1000 nm achieved an accuracy of 81.11 %, whereas an accuracy of 77 .17% with less computational load was achieved when SVM model was established on the optimal wavelengths selected. To determine the suitable altitude, speed of the sensor and minimum mapping unit, the altitude of the sensor was varied from 40cm to 150cm and the MMU was varied from 0.25 to lcm. The suggested altitude and speed of the sensor was 90cm and 10.5 cm/s. The model achieved an accuracy greater than 70%, whereas 0.25cm to 0.5cm achieved an accuracy less than 60%. The results of the study demonstrate that hyperspectral imaging system along with support vector machine is a promising tool for trace detection of explosives. |
| Year | 2020 |
| Type | Dissertation |
| School | School of Engineering and Technology |
| Department | Department of Information and Communications Technologies (DICT) |
| Academic Program/FoS | Remote Sensing (RS) |
| Chairperson(s) | Sarawut Ninsawat ; Nakamura Tai |
| Examination Committee(s) | Miyazaki, Hiroyuki ;Hornyak, Louis |
| Scholarship Donor(s) | Government of Japan |
| Degree | Thesis (Ph.D.) - Asian Institute of Technology, 2020 |