A hybrid system equipped with multi sensors as perception sources is necessary when robot executes complex tasks. There are two most adopted camera configurations for robot, including eye-in-hand and eye-to-hand, which has different advantages. It is crucial to determine the relationship between the robot base, the robot hand, the fixed camera and the camera installed at the end of the robot, that is, the robot-world and hand-eye calibration of the hybrid system. This study is focused on the joint treatment of the robot-world and hand-eye transformations. A linear method based on Kronecker product and singular value decomposition is proposed via two models. The two models involve the matrix equation AXB=YC and the enhanced dual equations. The method estimated the rotational and translational parts individually, which reduced the negative influence from the translational part. Their feasibility and effectiveness were further tested using simulated and real experimental datasets. The results obtained suggest that both of them yielded the good results in accuracy.