An Algorithm of Window Choice in Depth by Statistics Feature in GPR Target Detection and Location

申家全,闫怀志,胡昌振

2010-01-01

Abstract:Target detection and location is important in the application of the ground penetrating radar.Target location is based on the result of the detection,which in turn depends on the hyperbolas of the ground penetration radar.Many algorithms for this purpose involve a large amout of computational effort,such as SAR and overlay amplitude-velocity algorithm.They include an algorithm on windowstatistics to consider noises and non-targets data.At the same time,Hough transform algorithm has also to be used.Based on the similarity of time samples of ground penetrating radar and the fact that the possible target lies in some area,an algorithm is proposed in this paper for ground penetrating radar target detection and location.First,the algorithm considers energy and variance as statistics features of one-dimensional data in depth,to determine the probable depth of target signals by the right threshold.Then the energy variant is used as the statistics feature of an A-scan in the above special depth window by the threshold.Finally,the target location is determined.Because it only uses a part of A-scan data without considering the non-targets data,the location is more accurately determined than the algorithm that uses the whole A-scan data.Moreover,the algorithm is more effective than above mentioned other algorithms.The result of the test shows that the algorithm is very effective in the fast target detection and location.