Texture Image Analysis of Metallography: Automatic Estimating Grade of Spherular Pearlite Using Dempster-Shafer Theory

An algorithm based on Dempster-Shafer evidence theory with discernment frame segmentation is proposed for texture image analysis, which applies to automatic gradation of spherular pearlite about 15CrMo. Image enhancement, segmentation and feature extraction is implemented first to form the feature space, which includes the fractal dimension, energy and entropy. The feature information is fused using the proposed algorithm. The experiment demonstrates that the algorithm applied to the case with both high accuracy and efficiency