In this paper, we present a novel approach to the problem of curve alignment, which measures and matches the similarity of two curves. Our method extracts curve features with high priorities given to salient and general features, and therefore leads to a satisfied result that meets human being?s perceptions. We investigate the bimorphism in the level of subsegment correspondences to ensure symmetric mapping. A solution to closed curve alignment is also given. Finally, the coarse-to-fine aligning algorithm is introduced to match curves under different resolutions. The experiments showed satisfying and approving results, that our approach can capture the curve features and align them properly.