Statistical Modeling of Surface Roughness and its Estimation using Neural Network

Surface roughness, is a measure of surface quality is one of the specified requirements in a machining process. The machine vision applications have been carried out many researches in industries, as they have the benefit of being non-contact and speedy process than contact methods. In machine vision, is possible to analyze and determine the area of the surface, in which machine vision information will assist sensors to make intelligent decision on the applications. In this work, surface roughness estimation has been done by machine vision system. The extraction of features for the enhanced images is in spatial frequency domain done with the facilitate of Fourier Transform and Wavelet Transform. A neural network (NN) is trained with feature extracted values as input acquired from wavelet transform and examined to obtain Rt as output. The estimated surface roughness parameter (Rt) based on NN, which is compared with the Rt values from Stylus method is obtained as results.