Remote sensing is usually used in military to detect camouflaged object and geophysical prospecting. This project uses hyperspectral remote sensing to detect landmine. Because hyperspectral imaging has much more bands than multispectral imaging, it has more spectral information. We detect landmine according to the features of its spectral signatures. Landmine accounts for a small proportion of the entire image and concealed target can’t be detected in visible light, so using hypersepctral remote sensing to detect landmine.
The data of this project are collected by our own. The instruments include halogen light source, spectrometer, optical fiber, moving platform and landmine (replace landmine with battery). We change lots of scanning methods in
the process of data collection. Finally, we find a method that is accurate, full automatic and more time-saving than the one in the beginning of research. We compare the differences between fluorescent and halogen light, and the influences about the integration time and band selection by VD.
The bands of the hyperspectral imaging are numerous and its data size is big. Therefore, we need to reduce data size without losing important information. Using OSP algorithm to remove interference and to detect landmine. We also compare the results and differences among RX, CEM and OSP algorithm in landmine detection.
Comments