Recognition of Persian handwritten digits using image profiles of multiple orientations

In this paper a new approach for recognition of Persian (Arabic) handwritten digits is presented. This method utilizes the outer profiles of the digit image that are calculated at multiple orientations, as the main feature. Furthermore, the crossing counts and projection histograms of the image are used as complementary features. Similar to the profile features, these features are also calculated at multiple orientations. In the classification stage of our proposed method the support vector machines are applied. Evaluating the proposed system with approximately 4000 test samples the recognition rate of 99.57% is achieved. (C) 2004 Elsevier B.V. All rights reserved.