Abstract: In different spatial data set, corresponding enties have different space pattern of manifestation, while identification of multi-source heterogeneous data sets of the same entity is the fundamental problem in spatial data integration and various kinds of its applications. Integrating geospatial data from different sources is the most important way to improve the quality of GIS data, while identification of the same object is the prerequisite for integration and analysis of spatial data. In this paper, on the basis of the shape characters, we classify the complex linear features into the simple and the complex linear features, In terms of matching of curve features, there are still some shortcomings in existing methods, this is the reason that the paper puts forward complex curve features of the Fréchet distance matching method. Based on geometry and topology characteristics of curve features, the method first obtains the matching candidates, and then realizes the simplification of features on the basis of Fréchet distance and simplified method. Finally, by introducing triads of simplified features to store the attribute information of the complex linear element, the paper proposed an improved method based on the Fréchet distance and the triads information of the complex linear feature to accomplish the detection of different types of corresponding pairs and to realize complex curve features matching. The test results show that the proposed method not only can efficiently solve the matching problems of curve features, but also can effectively identify 1:0, 1:N and M:N matching pairs.