Digital image processing algorithm for industrial on-site roughness evaluation in Ti-alloy machining

The surface texture is normally observed after the machining process, but nowadays it is important to use on-site analysis to improve the process automatically via smart processing. This study introduces a contactless roughness inspection method employing digital image processing on Ti6Al4V samples in turning using three different feed. Texture analysis with grey-level co-occurrence matrix (GLCM) extracted features that were correlated with the arithmetic average roughness (R-a), leading to the establishment of predictive models. The study encompassed diverse image testing, incorporating variations in resolution and brightness distributions. It was found that the pixel pair spacing (PPS) in GLCM analysis was influenced by the image resolution and feed rate. The predictive models developed with high-quality images, i.e., higher resolution and better brightness distribution, yielded similar results to those created using lower-quality images.