Unfolding wheels are critical equipment in steel ball surface defect detector

which affects the detection precision directly. Currently

unfolding wheel can only be fetched through importing. To address the problem

this paper applies reverse engineering techniques on modeling of unfolding wheel. First

a three-dimension raster scanner is used to perform a digital scanning on the unfolding wheel

resulting in an ＇initial digital model. Then

a hybrid modeling method using spline and freeform surface feature is proposed based on the digital model. Next

part structure parameters and the characteristic parameters for working surface of the unfolding wheel are extracted ac- cording to model of unfolding wheel. Finally

curve fitting is performed on the section point cloud using MatLab

the working surface linear equation is obtained. The result shows that the efficacy of the high-accuracy modeling method proposed in this paper

which serves as a theoretical base for machining the unfolding wheel.