Intelligent recognition for surface roughness based on microscopic image texture characters

Aiming at problems that the traditional measurement of surface roughness is complex and the accuracy is lower caused by man-made, this paper researched an intelligent measurement method based on microscopic image texture characters. The original microscopic images are acquired on the microscope, after filtered and histogram equalized, six texture characters such as second-order distance, contrast, correlation, entropy, anti-difference distance are extracted from the gray images with theory of GLCM and these parameters are changed regular. At last, using pattern recognition theory of nerve network, we will get the surface roughness value of the workpiece. The results of experiments show that this method can identify the value of surface roughness, and it provides a new approach for the measurement of surface roughness.