DEVELOPMENT OF STRATEGIES FOR SUB-SURFACE OBJECTS AND LANDMINE DETECTION

Abstract

The immense operational and intelligence potential of remote sensing continues to entice military strategists around the world. Microwave remote sensing, for example, has the ability to detect underground landmines, particularly along the country's western borders, which are covered in vast deserts devoid of vegetation. In the current study, this issue has been investigated. Landmines are small explosive devices that are planted at shallow depths to kill or incapacitate an unsuspecting opponent. Microwave frequencies in the X-band (10 GHz frequency and 3 cm wavelength) may have adequate penetration and resolution for landmine detection at shallow depths where landmines are buried, but the backscattered image from these shallow buried landmines is severely cluttered, and mine feature extraction remains the main problem in landmine detection. Several signal and image processing approaches that have been proposed to handle the problem either have functional limits or produce a large number of false alarms. The major focus of this research is to look into the methodologies and processes for detecting subsurface and surface targets utilizing microwave data. Landmine detection up to a depth of 10cm has been tested in the lab and in the field using microwave X-band frequency (10Ghz, 3cm). In the laboratory, data was collected using a dummy antitank landmine (without explosives) buried in dry smooth sand at various depths, as well as a live antitank mine (in the field). Raw data is processed using two separate methods: full image processing and local window processing. Data pre-treatment entails a series of image processing processes prior to segmentation utilising Otsu's and maximum entropy based thresholding. Local window-based processing is found to highlight mine-like and non-mine-like characteristics more precisely than whole image processing, allowing for straightforward segmentation utilising two thresholding methods. iv The usage of local window processing, on the other hand, may result in a longer processing time. The two thresholding methods have a minor difference in performance, with the maximum entropy-based method performing somewhat better in segmenting mine-like features with low variance from surrounding clutter. The findings were confirmed using the known location of the mine that was used in the trials. The presence of a landmine is indicated by a detection figure in the range of 30-80. The backscattered electric field was theoretically calculated using Daniels' suggested electromagnetic model. The backscattered electrical field detected was acquired from a segmented suspected landmine-containing zone.