In the context of space surveillance applications, we propose a pose estimation approach for space objects based on contour feature matching. The proposed method combines global appearance features and local contour features to perform coarse-to-fine pose estimation, and adopts a voting strategy based on geometric errors to obtain robust pose results. It applies to texture-less space objects without relying on priori of specific shapes. Moreover, we design rendering simulation pipeline for space objects, which considers imaging characters of space surveillance and multiple degradation factors. Using this simulation pipeline, we construct a space object image dataset to provide support for vision-based research. Experiments analyze the adaptability of proposed pose estimation methods to a variety of complex imaging conditions in space environment, and demonstrate our advantages in terms of pose estimation accuracy and computational efficiency.