Complex Surface Inspection Based on the Minimum Zone Criterion

Although there have been many studies in free-form surface inspection, it remains a difficult problem in application because of different inspection results resulted from various registration methods. Accurate registration is vital in precision manufacturing. Otherwise, rejection of some qualified parts would occur. A new registration method is presented to find the spatial transformation with which the surface error range is minimized. A smooth function is used to approximate uniformly the objective function of the registration to deal with the computational intractability. The limited-memory Quasi-Newton algorithm is prevalent for large-scale optimization problem. It is a differentiable unconstrained optimization method, which only need to provide the function and gradient values of the objective function. The proposed method is easily implemented and works well for dense measurement points. Experiments justify the superiority of the proposed algorithm.