This paper investigates the hand-eye calibration and points registration methods based on Cartesian coordinate’s transformation in armed robot inspection system. Different from traditional imaged based approach, the sphere radius of the inspection probe is considered; the hand-eye calibration and 3D points’ registration are both derived as constrained optimization problem. Matrix transformation of parameter separation is presented to get the approximate resolution. Finally, iterative cycles are provided to calculate the nearest points on designed surface corresponding to the inspection points. Experiment results illustrated the application of these ideas and demonstrated the validity of the proposed framework for inspection robots.