Multicamera systems have many advantages and are widely used. However, many situations require camera parameters that are more accurate than those that are currently available. A new algorithm is proposed to improve the accuracy and consistency of these systems by adjusting the camera parameters. The algorithm assumes that the distribution of the measured point positions follows the Gaussian mixture model. Based on this model, point positions in space are estimated, and new camera parameters are computed from the estimation. A metric is defined to describe the difference between the newly computed and precalibrated camera parameters, following which the parameters are adjusted by minimizing this difference. Finally, the validity of the algorithm is confirmed by conducting experiments. Two indicators that describe the accuracy and consistency are defined and applied to analyze the experimental data.