Evaluation of surface roughness standards applying Haralick parameters and Artificial Neural Networks

This paper presents a methodology for roughness analysis based on surface characteristics of images obtained from optical and electronic microscopes. Texture analysis has been widely used in different fields, such as medical image analysis, visual interpretation of remote sensing images, image search and industrial quality inspection of manufactured products, which is the focus of this paper. We present a novel method of analysis based on texture characteristics. The Haralick descriptors are used to describe the surface texture and to classify its roughness. The primary roughness standards are evaluated and classified according to several features considering these descriptors. The set of values is the input of a multilayer perceptron artificial neural network. The obtained results show that it can be possible to advance this methodology in order to develop a roughness digital measurement system for the evaluation of surface ending process.