Linear pose estimation with a monocular vision system

The algorithm that determines the target position and orientation according to the image coordinates of reference points is very important in pose estimate technique. Aiming at this issue, a nonlinear pose estimate mathematic model is established on the basis of distance factor. With algebraic transformational method, the nonlinear mathematic model is transformed into a linear nonhomogeneous equation system. The product relationship between the general solutions of the equation system is obtained according to the variable relation in the algebraic transformation. Considering the unique solution of pose estimate problem with four coplanar reference points, using the algebraic transformational method again, the linear pose estimation is realized. Both simulation and real data experiments are used to examine the accuracy and stability of the pose estimate algorithm. Experiment result shows that the proposed algorithm can reduce the interference of image noise and improve pose estimate precision.