Fusion at the feature level for person verification based on off line handwriting and signature

The off line identification of the handwriting as of signature comes under the field of biometrics. The context of use is in particular in the banking and legal fields. Within this framework problems particularly of imitation and falsification are often met. This paper presents an approach of personal identification based on the fusion of two off line modalities: handwritten signature and handwriting. The method is based especially on exploration of textual characteristics extracted using the wavelets. Identification is ensured by a support vector machines (SVM) classifier. Encouraging overall performances are recorded.