This paper presents a new offline signature recognition system based on Modular Neural Networks (MNN) and fuzzy inference system. The proposed MNN consists of three different modules, each using different image features as input, these are: edge detection, curvelet transform, and the Hough transform. The Mamdani fuzzy inference system is then used to combine the outputs from each of these modules. The experimental results obtained by using a data base of 30 individuals’ signatures show that the proposed modular architecture can achieve very high 96.6% recognition accuracy with a small test set of 60 images. Therefore, we can conclude that modular architectures and more specifically our proposed system can provide a suitable platform to build a signature recognition system.

Keywords