Evaluation of Background Subtraction and Pattern Recognition Methods for Classification of GPR B-Scan Data

In this study, the performance comparison of background subtraction, feature extraction and machine learning based classification methods on GPR (Ground Penetrating Radar) data obtained from buried objects was made. Filtering, pre-processing and post-processing techniques were used, and after making the buried objects more distinct, the data was classified using machine learning methods. The developed methods were tested on a real data set consisting of 1040 data and 6 different classes and the performance comparison of the methods was made.