In this paper, machine vision technology was adopted to ass’s surface roughness of the parts machined by the milling process. Machine vision allows for the assessment of surface roughness without touching or scratching the surface. About 32 samples were machined by the milling process under various cutting parameters and images of the machined surfaces were captured using CCD camera. Consequently, widely used conventional average surface roughness ‘Ra’ is measured using surfcorder (A stylus based instrument). The images are processed using various image processing techniques in MATLAB, to ease the extraction of surface roughness features. The average of the variances of randomly selected columns from the pixel matrix of the processed gray scale indexed image were calculated and taken surface roughness feature ‘F’. Similarly, one more feature ‘E’ called entropy was derived from the processed image. Both the features were found to be fairly in good correlation with the conventionally measured surface roughness Ra. A simple multiple regression model was also generated, which better represented the relationship between the derived features and the surface roughness Ra.
Volume 11 | 07-Special Issue
Pages: 500-509