QUALITIVE SURFACE ROUGHNESS EVALUATION USING HARALICK FEATURES AND WAVELET TRANSFORM

The research presented in this paper is aimed at the development of an automated vision system for classification of surface finish in turning parts. Haralick features in the frequency domain have been used to characterize finish roughness of metallic parts by means of the wavelet transform. First, the wavelet transform was applied to two sets of images, one belonging to parts with low roughness and other belonging to parts with high roughness. Haralick features were worked out for the image description. Then, a classification was performed using knn. Others simple and based on moment features were also calculated to compare the results against Haralick features.