Joint position and joint angle are the basic kinematic parameters for the quantitative evaluation of upper limb motion ability, which are widely applied to calculate extremity reachable workspace and range of motion (ROM). The traditional measurement methods cannot meet the clinical requirements in accuracy, stability and real-time. This paper proposes a novel algorithm for measuring the kinematic parameters of upper limb by combining the kinematic chain method with the vector projection method. Firstly, the human upper limb motion model is simplified to a 7 degrees of freedom (DOF) motion model according to the anatomical principle, and the rigid body group is used to collect the upper limb kinematic parameters to realize the real-time tracking of trajectory and angle of the shoulder, elbow and wrist joints. Finally, a healthy volunteer experiment is designed, the results show that the algorithm can provide real-time and quantitative upper limb kinematic parameters, which can meet the data requirements of quantitative evaluation of upper limbs.