It can improve the service quality of university library by matching the feature points of the users information. For the efficient feature points matching of user information

it is necessary to give the threshold of different types of resources correlation degree

and to construct a new feature point matching feature expression of user information. In the traditional method

the matching two step projection matrix is formed by the constraint condition of the user information

but the construction of the characteristic expression is ignored

which leads to the low matching accuracy.This paper proposes a method to match the feature points of user information based on probability generation. In the fusion probability assignment generation theory to set up a user information probability generation model

the latent variables of unknown type defines the user information for the model

calculate the joint probability distribution

distribution of resources of each type of access to a posteriori estimates

given different types of resources associated with the measurement of the threshold

construct the matching feature expression

calculate the most the characteristics of short distance between adjacent types of user information

the user information matching feature points of University library. The simulation results show that the proposed method can improve the accuracy of the feature points matching

which can lay a solid foundation for improving the service quality of University library.