Aiming at the problem of long time, large matching error and many registration models for cooperative object based on orb feature matching, a global cooperative object pose calculation method based on sparse optical flow method and ray triangulation method is proposed. Based on the sparse optical flow method, the two-dimensional feature points are tracked, the points with bad matching are eliminated, and the pose calculation is realized by pnp algorithm. If the matching points are less, the feature points in the current frame are extracted by ray triangulation method and the world coordinates are estimated to improve the robustness of the algorithm. When the camera parallax changes too much, in order to prevent the accumulation of new feature point error, the feature matching is carried out in the original registration model, and then the current pose is calculated. Experimental results show that the speed of pose calculation is 6 times higher than that of orb feature matching algorithm, the error of feature matching is reduced by 19%, and the registration of model is reduced by more than 50%. And the algorithm has good applicability to illumination change, angle change and target surface similarity.