Humanitarian Microwave Imaging Enhancement and Classification of Shallowly Buried Objects

Mostafa El Saadouny, Jan Barowski, Ilona Rolfes

10th IEEE Annual Information Technology, Electronics and Mobile Communication Conference (IEMCON 2019), pp. 394-397, doi: 10.1109/IEMCON.2019.8936165, Vancouver, Canada, Oct 17-19, 2019


The ground penetrating radar (GPR) is one of the promising tools for investigating the shallowly buried objects. This paper aims to solve the problem of the strong clutter reflections associated with the GPR images, and also classify the detected targets to differentiate between different buried objects. First, the implemented algorithm tends to remove the strong background bounce and clutter originated in the non-homogenous region under test by combining two levels of clutter reduction techniques. The first level consists of applying the Moving Average Background Subtraction (MA-BS) method to remove the strong ground reflections and emphasize the target reflection. The second level consists of a combined algorithm which implements the DC-offset removal and the Subtract and Weight (SaW) method for further reduction of the clutter level. The theoretical and mathematical formulation is presented and the acquired results have been investigated for evaluating the implemented method. After reducing the clutter to the lowest possible level and acquiring the images of the buried objects, a K- nearest neighbour (KNN) classifier has been implemented to classify the GPR images. The KNN has been trained to solve this classification problem by a large number of GPR images. After preparing the training data, the data with the corresponding labels have been used for training the KNN and afterwards, the algorithm has been tested and the obtained results showed a good accuracy in objects classification.

[IEEE Library]