Gradient of images is an effective discriminative feature, widely used in pattern recognition applications. In this work the twelve directional codes depending on the gradient direction is coupled with a statistical classifier for designing an offline recognition system for handwritten isolated Malayalam characters. Preprocessed character images are decomposed into sub-images using the Fixed Meshing strategy and the twelve directional codes are extracted to form the feature vector. Classification has been carried out by implementing the Modified Quadratic Discriminant function (MQDF), a successful statistical approach for Handwritten Character Recognition. We obtained 95.42% accuracy, and the experimental result shows that the approach provides better results. Compared to QDF, MQDF improves the classification performance by more than 10%, reduces the computation cost and also provides dimensionality reduction to a larger extent. A database of 19,800 handwritten Malayalam character samples was used for the experiment.

1. Bindu S Moni, Raju G, “Meshing and Normalized Vector Distance from Centroid for Handwritten Malayalam Character Recognition”, 2nd Int. National Conference on Signal and Image Processing (ICSIP – 2009), Mysore, 2009, pp 398 – 403, Aug 12 – 14.
2. Bunke H., Wang P.S.P., Handbook of Character Recognition and Document Analysis, World Scientific, 1997.
3. Cheng-Lin Liu, “Normalization-Cooperated Gradient Feature Extraction for Handwritten Character Recognition”, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 29, no. 8, August 2007.
4. Nagy G, “Twenty Years of Document Analysis in PAMI”, IEEE Trans. On PAMI, Vol 22(1), pp 38 – 61, 2000.
5. Plamondan R., Srihari S.N., “Online and Offline Handwriting Recognition: A comprehensive Survey”, IEEE Trans. On PAMI, Vol 22(1) pp 63 – 84, 2000.
6. Rajasekharadhya S.V., Vanaja Rajan P., “Efficient Zone Based Feature Extraction Algorithm for Handwritten numeral Recognition of four Popular South Indian Scripts”, J. of Theoretical and applied Information technology, pp1171 – 1181, 2008.
7. Raju G,“Recognition of Unconstrained Handwritten Malayalam Characters Using Zero-crossing of Wavelet Coefficients”, Proc. of 14th International Conference on Advanced Computing and Communications, 2006, pp 217 – 221.
8. Raju G, “Wavelet Transform and Projection Profiles in Handwritten Character Recognition – A Performance Analysis”, Proc. 16th International Conference on Advanced computing and Communication (ADCOM)*), 2008, pp 309-313.
9. Raju G, Bindu S Moni., “Global Elastic Meshing for Handwritten Malayalam Character Recognition”, Proc. of the National Conference on Computational Science and Engineering - NCCSE 2009, Cochin, Kerala, February, 2009, pp 10 – 14.
10. Raju G, Bindu S. Moni, “Global and Local Elastic Meshing for Handwritten Malayalam character Recognition”, Int. J. of Computers, Information Technology and Engineering, Vol. 3, No. 1, January - June 2009, pp. 149 – 153.
11. Renju John, Raju G and Guru D. S., “1D Wavelet Transform of Projection Profiles for Isolated Handwritten Character Recognition”, Proc. of ICCIMA07, Sivakasi, 2007, Vol 2, 481-485, Dec 13-15.
12. Srihari S.N., Yang X. and Ball G.R., “Offline Chinese Handwriting Recognition: an assessment of current Technology”, Front. Comput. Sci, China, Vol. 1(2), pp. 137-155, 2007.
13. Trier D, Jain A.K. and Taxt T, “Feature Extraction Methods for Character Recognition - A survey”, Pattern Recognition, Vol 29, 4, pp 641 – 662, 1996.
14. Bindu S Moni, G Raju, “Multiple MLP Classifiers for Handwritten Malayalam Character Recognition”, Proc. of the Int. National Conference on Mathematical and Computational Models: Recent Trends- ICMCM-09, Coimbatore, 2009, pp 349-354, Dec 21-23.
15. Binu P. Chacko, Babu Anto P., “Discrete curve Evolution Based Skeleton pruning for character recognition”, Proc. of IEEE International Conf. in Pattern Recognition, 402 – 405, Feb 2009.
16. Bindu S Moni, G Raju, “Quadratic Classifier for Handwritten Malayalam Character Recognition”, Proc. of the U G C sponsored National Conference on Soft Computing, organized by Dept. of Computer Applications, Marian College Kuttikkanam, Kerala, Jan 20 – 22, 2010, pp. 59 – 68, ISBN: 978-81-908520-1-2.
17. Hailong Liu and Xiaoqing Ding, “Handwritten Character Recognition Using Gradient Feature and Quadratic Classifier with Multiple Discrimination Schemes”, Proc. of the 2005 Eight Int. Conference on Document Analysis and Recognition (ICDAR’05).
18. Bindu S Moni, G Raju, “Study on different Meshing Techniques and Normalized Vector Distances for Handwritten Malayalam Character Recognition”, Int. Journal of Engineering Research and Industrial Applications (IJERIA), Vol. 3, No.1 (Feb 2010), pp 181 – 195, ISSN 0974-1518.
19. Bindu S Moni, G Raju, “ Runlength Counting for Handwritten Malayalam Character Recognition”, Proc. of the AICTE sponsored Int. National Conference, ICMCM 2010, organized by MACFAST, Thiruvalla June 17-19, 2010.
20. Weipeng Zhang, Yuan Yan Tang, Yun Xue, “Handwritten Character Recognition Using Combined Gradient and Wavelet Feature”,1-4244-0605-6/06/$20.00 ©2006 IEEE.
21. Dayashankar Singh, Sanjay Kr. Singh, Dr. (Mrs.) Maitreyee Dutta, “Hand Written Character Recognition Using Twelve Directional Feature Input and Neural Network”, ©2010 International Journal of Computer Applications (0975 – 8887) Volume 1 – No. 3.
22. Bindu S Moni, G Raju “ Modified Quadratic Classifier And Normalised Vector Distance For Handwrittan Malayalam Character Recognition”, Proc. of the Int. National Conference on Emerging Trends in Mathematics and Computer Applications – ICETMCA 2010, organized by MEPCO Schlenk Engg. College, Sivakasi, , Dec 16-18, 2010, pp 356 – 360, ISBN : 978 – 81 – 8424 – 649 - 0.

Fixed Meshing Gradient Features Handwritten Character Recognition Quadratic Classifiers