Résumé:
In this manuscript, an automatic system for the analysis andclassification of ancient manuscripts is introduced. The proposed system comprises two main steps: feature extraction and classification (writer identification). In the first step, distributions run-lengths, distributions of contour directions, distributions of contour hinges are extracted from ancient handwritten documents. In the second step, we have used support vector machines (SVM) with one against all strategy for classification. The experiments are carried out on a dataset which includes 735 ancient handwritten documents. The highest writer identification rate is achieved by the use of SVM classifier with edge-hinge distributions