2014 · 16 citations · 6 references
ComputingComputer ScienceEarth SciencesComputer VisionFeature EngineeringMachine LearningGround-penetrating RadarImage ClassificationKernel MethodsObject DetectionObject RecognitionSupport Vector MachinesBlob DetectionBuried Object DetectionGeophysical ExplorationGround Penetrating Radar ImagingLandmine DetectionSingular Value DecompositionSpeeded-up Robust Features (Surf)Support Vector Machine Classification
Detection and classification of an object is of growing concern in many application areas. Ground Penetrating Radar (GPR) data has been widely used in fields like military, archeology, and geophysical exploration and many such applications. In this paper, we present a combination of Singular Value Decomposition (SVD) approach and Blob detector for detection of buried objects from ground penetrating radar (GPR) data. Classification is done using Speeded Up Robust Feature (SURF) and Support Vector Machine (SVM) classifiers for landmine identification. 3-Dimensional data is collected using GPR Vehicle Mounted system. Buried objects are detected using the proposed SVD and blob detector approach. The located buried objects are then classified to discriminate between landmines and other objects. Performance of the proposed system was evaluated for different kernels of SVM classifiers for a number of datasets collected for different targets and soil conditions.
6
Speeded-Up Robust Features (SURF)
Herbert Bay, Andreas Ess, Tinne Tuytelaars et al. · Computer Vision and Image Understanding · 2008 · 13.4K citations
Ground Penetrating Radar Fundamentals
Jeffrey J. Daniels · 2000 · 74 citations