A novel calibration algorithm is presented to integrate structured-light vision sensor into coordinate measuring machine.The system modeling is decomposed into two interactive parts:establishing the model of structured-light vision sensor and calculating the transformation between skewed vision sensor and CMM Cartesian world frame. A B-spline function is applied to approximate the mapping model of structuredlight vision sensor in the skewed sensor frame. Given initial coordinate transformation matrix, sample points for establishing B-spline function in the skewed sensor frame is calculated and B-spline function can be obtained by an iterative adjustment approach. Then, a nonlinear constrainted optimization algorithm is used to adjust the coordinate transformation matrix and B-spline function model of vision sensor is refreshed. The above procedure is repeated until an stable optimal transformation matrix is obtained. an experiment is presented.