Experimental approach of quality enhancement for image acquisition of ground penetrating radar

Abstract

Due to the time consumption required to perform data acquisition, sometimes the image quality of scanning process for ground penetrating radar (GPR) needs to be sacrificed to obtain the result. Hence, the image becomes low-quality for further processing. In this paper, an experimental approach is performed to enhance the image quality for GPR acquisition. A model of M16 riffle model is used as the object of investigation which is buried in the GPR test range at the depth of around 350mm under the sand surface. The scanning process is then run out to obtain the data of buried object. The experimental data is processed into A-scan, B-scan, and C-scan. Image interpolation is performed to the C-scan image by using 3 different methods, i.e. nearest neighbor method, linear method, and cubical method. The peak signal-to-noise ratio (PSNR) of interpolated image from each method is analyzed. It shows that the linear interpolation method gives the highest PSNR, i.e. 50.73dB, over other methods, i.e. 48.15dB and 49.19dB for nearest neighbor and cubical interpolation methods, respectively.