This work presents a filter algorithm for more accurate 3D pose estimation of a maneuvering target. First, by analyzing the error propagation from the 2D images to the target pose, linear measurement equations are derived. Then, two filter schemes are proposed. The first filter uses the maneuver detection technique, in which two detectors fit for fast and slow maneuvers are deduced respectively, and limited memory filtering is adopted for maneuver correction. In the second filter, a robust dynamic model is constructed by the numerical differentiation technique, and an adaptively estimated fading factor is imported to restrain the filter divergence. Finally, the two filter modules are combined for more accurate estimation. Superior to previous approaches that were limited to the slowly and smoothly moving target, this algorithm is applicable for a maneuvering target that acts in an unknown manner. Simulation results show the capacity of this algorithm.