Trademark classification by shape using ensemble of RBFNNs

Identification of similar trademarks is important in trademark registration. Shape feature could intuitively and effectively describes an object in a given image. Therefore, shape feature plays an important role in content-based image retrieval (CBIR) systems. The shape feature is particularly suitable for trademark image retrieval (TIR) systems. In this paper, we propose an effective solution for TIR by using Hu's invariant moments and an ensemble of Radial Basis Function Neural Networks (RBFNN) trained via a minimization of the Localized Generalization Error Model (L-GEM). The proposed method outperforms TIR with similarity measure based on Euclidean distance.