DEVELOPMENT OF DETECTION AND IDENTIFICATION TECHNIQUES FOR GROUND PENETRATING RADAR

Abstract

This dissertation deals with the development of detection and identification techniques for a GPR. An SFCW GPR is used for the detection and identification of several targets like pipes and cables. A background subtraction using reference background data has been implemented along with several preprocessing techniques. A fast back-projection algorithm (BP) was implemented. Several filtering techniques were used for the removal of noise from the image and statistical technique Principal Component Analysis (PCA) has been implemented for the reduction of clutter from the images. A mean-deviation based thresholding is implemented for the segmentation of the image. After the detection of the targets, the time-frequency representation of one-dimensional signal is obtained by applying Wigner Ville Distribution. This time-frequency matrix is decomposed using Singular Value Decomposition into three matrices and are used for the feature extraction and these features can be used for classification for the identification of materials.