To obtain ultra-high spatial resolution image, a kind of non-conventional fiber bundle, whose input was line array while output was plane array, was designed. The coordinate table of each fiber in input and output terminals must be ascertained when synthesizing object images by using the scanning-imaging system of the fiber bundle. However, limited by manufacture technology, the fiber bundle output structure was not strictly aligned. An algorithm based on Kalman Filter, aiming at the coordinate errors, was proposed in this paper. Different initial searching step was given to inner layer and intra layer respectively, by searching fiber center’s position one by one in the output terminal, under the principle of minimum mean square error. The coordinate table of the output was established, and the position mapping relationship between output and input was built. Comparing the termination margin and marking the searched fiber were the two ways to improve the robustness of the algorithm. Z-shaped path improves the efficiency of the algorithm. The results show that the coordinates obtained by the proposed algorithm can demarcate the position of the fiber and the object image was restored by extracting the information of the scanning imaging system coupled by the bundle.