Abstract: The technique of 3D detection based on structured light is applied to 3D surface defect detection of steel rails during production. Four linear lasers and eight CCD area-array cameras are installed to detect the four sides of steel rails. Images of laser strips captured by CCD cameras are processed to get the depth variation value of steel rail surface with steps of extraction of strip centers, correction of strip centers, subtraction of strip centers by benchmark, and so on. Distribution of depth variations along the direction of length and height of steel rails is plotted as distribution map of depth variations, which is a 2D gray image. Areas of defects are detected by way of image recognition, thus realizing automatic defect detection. On-line industrial application of the technique shows that the maximal detection speed is 1.5 m/s and the resolution of depth detection is 0.2 mm.

Key words: 3D detection, Steel rail, Surface defect