Aiming at the problem that the current rolling contact fatigue test fails to detect the three-dimensional（ 3 D） morphological characteristics of defects

a 3 D reconstruction method of fatigue roller surface based on point cloud data is proposed. Firstly

a 3 D laser scanner is used to obtain point cloud raw data on the surface of the fatigue roller; gradient segmentation method is used to achieve the segmentation of fatigue contact surface; the KdTree algorithm in the Statistical Outlier Removal filter is used to remove outlier noise; and the curling of the tiled point cloud is realized; finally

greedy triangulation and hole restoration are performed on the curled point cloud.The experimental results show that the segmentation precision of the fatigue contact surface is above 97. 7 %

error rate of the point cloud curling is 0. 09 %

and the maximum deviation of the fatigue roller surface after reconstruction is 0. 019 9 mm

which can provide basis for analyzing on contact fatigue performance of roller specimen under simulated working conditions.