Abstract: To improve the accuracy and stability of joint mobility measurement, a measurement method based on variance adaptive weighted fusion is proposed. Multi- camera was used to collect bone joint nodes, a quartz- point method was added under the condition of bone length constraints to replace the group points, and the variance and total difference of the skeletal joints under different camera coordinates were calculated. By adaptively determining the optimal fusion weights based on the variance of different cameras and their proportion to the total variance, the fusion was completed after meeting the error conditions. Experimental results show that compared with single- camera, weighted average fusion, backpropagation neural network fusion and Kalman filter fusion, the accuracy and stability of the measurement results are significantly improved, and the maximum measurement error of the proposed method is controlled within 2.3°, which verifies the feasibility of the proposed method.