Recently, the triangle algorithm has become the most widely used star identification algorithm because of its simplicity and convenience, where the magnitude information plays a key role in the construction of star map features. However, in practice, the magnitude information of the observed star map is often difficult to use, because they might contain errors or be lost in some worst cases. To solve this problem, we proposed a multi-view double-triangle algorithm for star identification in this paper. This algorithm constructs double-triangle features of stars with the angle and distance information of star points. Moreover, to reduce the influence of noise interference on the identification accuracy of the model, we built multi-view double-triangle features for the observed star map to improve the robustness of the algorithm. Synthetic and real experiments show that our algorithm has a high identification accuracy of more than 98.4% in face of “false star” noises and “missing star” noises, and our algorithm is not affected by the focal length and the shooting angle of the star sensor. Moreover, the results also show that our algorithm has good robustness, short identification time and reduced storage costs, which could be beneficial in practice.