Off-line handwritten Arabic word recognition using SVMs with normalized poly kernel

Citation

Alalshekmubarak A, Hussain A & Wang Q (2012) Off-line handwritten Arabic word recognition using SVMs with normalized poly kernel. In: Huang T, Zeng Z, Li C & Leung C (eds.) Neural Information Processing: 19th International Conference, ICONIP 2012, Doha, Qatar, November 12-15, 2012, Proceedings, Part II. Lecture Notes in Computer Science, 7664. Berlin Heidelberg: Springer, pp. 85-91. http://link.springer.com/chapter/10.1007/978-3-642-34481-7_11#

Abstract
Handwriting recognition is a complicated process that many applications rely on, such as mail sorting, cheque processing, digitalisation and translation. The recognition of handwritten Arabic is still an ongoing challenge mainly due to the similarity among its letters and the variety of writing styles. In this paper, a novel approach is proposed that uses support vector machines (SVMs) with normalized poly kernel. The well-known Arabic handwritten database, IFN/ENIT-database, which contains 936 city names with more than 32,492 instances, is used to test the proposed system. The results of this novel approach are compared with the results of two different studies. The comparison shows that a higher accuracy rate is obtained using the proposed system.