Global Motion Estimation Algorithm Based on Pyramid Iteration

Aiming at the computational complexity and accuracy of the global motion estimation algorithm, a fast algorithm for global motion estimation is proposed. the algorithm eliminates the adverse effect of foreground object in global estimation and only fully search the particular regions in image to get initial optimal motion vector, then through the improved three-level pyramid multi-resolution model, obtain the optimal motion vector for global compensation. The simulations show that the algorithm increases the parameter estimation speed without loss of imaging quality and estimation accuracy, which is practicable.