Detection of tool condition from the turned surface images using an accurate grey level co-occurrence technique

With the advancement of digital image processing, tool condition monitoring using machine vision is gaining importance day by day. In this work, online acquisition of machined surface images has been done time to time and then those captured images were analysed using an improvised grey level co-occurrence matrix (GLCM) technique with appropriate pixel pair spacing (pps) or offset parameter. A novel technique has been used for choosing the appropriate pps for periodic texture images using power spectral density. Also the variation of texture descriptors, namely, contrast and homogeneity, obtained from GLCM of turned surface images have been studied with the variation of machining time along with surface roughness and tool wear at two different feed rates.

▶ Accurate GLCM analysis is dependent on the appropriate selection of pixel pair spacing (pps) value for periodic texture images. ▶ Automatic computation of appropriate pps value has been done. ▶ A study of the condition of the cutting tool using GLCM technique with optimum pps has been done. ▶ Contrast and homogeneity were two texture descriptors found to be suited, where contrast is the most suited descriptor. ▶ The selection of optimum pps value can be done for any periodic texture images