Point cloud registration in shoe glue spraying line

We studied the high-precision registration method with a pair of shoes sole and vamp point cloud to gain the automatic shoe glue spraying line. A new method was suggested in the case that the point cloud pair doesn´t overlap completely and the initial positions are always different. Firstly the singular value decomposition algorithm was used for the rough registration such that the direction of the point cloud pair was the same and the distance of the point cloud pair was close to each other. Then the iterative closest point algorithm was applied to the high-precision registration. Lastly the local deformation of the sole contour points was derived by Bezier curve algorithm according to the requirement of the actual production. Actual test results presented the effectively of the method that the requirement of the glue spraying production line was satisfied.