Abstract: In order to estimate the camera's small amplitude translational dither motion, various mismatching factors were analyzed, such as self-moving objects, unstable edge details and smooth areas. After summarizing the characteristics of various factors and the existing related basic theories, and the same time on the basis of the improved C-1BT transformation, the fast global translational motion parameters estimation based on the pre-selection process algorithm were proposed. Firstly, before the local motion vector estimation, the pre-selection process was used to keep some special sub-blocks, because they can better represent the global motion nature. Secondly, according to the characteristic of the pre-selection algorithm, the initial arm-length predictive part of the traditional adaptive rood pattern search method was improved to reduce the risk of falling into local optimum. At last, the optimum global motion parameters were obtained, with the translational motion model and the statistical principles. MATLAB simulation results show that the proposed method fast and robustness.