With the exponential growth of online multimedia, image copyright protection now confronts the issue of large scale.In this paper, we propose a feature fusion based learning method for image hashing, aiming to achieve efficient image copy protection.It first effectively utilizes the correlation between different feature models, and then generates compact fingerprints for image representation, which prevents huge semantic loss during the process of hashing.To accurately map image features into the Hamming space, an objective function of semantic loss is constructed and minimized, which combine the influence of both the neighborhood structure of feature data and mapping error.Therefore, our hashing method not only preserves the local structure of individual feature, but also globally considers the local structures for all the features to learn a group of hash functions.The experiment results show that the proposed method outperforms the state-of-the-art techniques in both accuracy and efficiency.