Estimation of Surface Roughness Using AFM Image Adopting 2D FFT and Wavelet Analysis

Surface discharge studies were carried out with silicone rubber added with 3wt% nano boron nitride composites. The study innovatively involves the frequency domain image analysis for estimating the nanoscale surface roughness of silicone rubber nanocomposites and thereby average surface roughness can be predicted using it as a texture feature. Five frequency domain parameters were extracted from two-dimensional surface morphological atomic force microscopy images. The two-dimensional Fast Fourier transform is used to extract the roughness features in the spatial frequency domain. The 2D frequency parameters and AFM roughness parameters are statistically correlated and correlate well with the experimental results. The prediction is high when the average power spectrum and central power spectrum percentage parameter of the AFM image are used for the regression model. Multi-Resolution Signal Decomposition (MRSD) approach is used to investigate the time-frequency features of surface roughness induced by surface discharges employing silicone rubber nanocomposites in an SF
6
ambiance at various gas pressures.