Off-line Text-independent Writer Identification Using a Mixture of Global and Local Features

The existing works on writer identification consider global feature or local feature, respectively, but not both. Actually, both of global and local features provide the useful information for writer identification. Hence, this paper proposes a method for writer identification by using a mixture of global feature and local feature. In implementation, we utilize 2-D Gabor transformation as the global feature and Local Binary Pattern (LBP) as the local feature for writer identification. The experiment results show that the combination of global and local feature outperforms the utilization of each single one.