Languages

(UNESCO / Japan Young Researchers' fellowships programme)

Intelligent Knowledge Acquisition and Extraction Techniques for Neural Expert Systems Based on Machine Learning Techniques

Summary of research carried out: 
Intelligent Knowledge Acquisition and Extraction Techniques for Neural Expert Systems Based on Machine Learning Techniques

Off-line unconstrained cursive handwriting has been a popular field of research for many decades. After forty years of intensive research, it still remains an open problem. The challenging nature of handwriting recognition has attracted researchers from industry and academia. The commercial sector has shown significant interest in handwriting recognition research, owing to the large number of applications such as bank cheque processing, tax form processing, admission form processing, and so on. In recent years, techniques for recognizing handwriting have become more sophisticated in order to deal with real-world situations and to increase recognition rates. Preprocessing techniques play a vital role in increasing the recognition rate of unconstrained off-line cursive handwriting. This research reviewed all aspects of preprocessing techniques, including line removal, slant correction, skew correction and core-zone detection. Segmentation of cursive handwriting is also a crucial step in the entire process, which is the focus of this research and which is examined thoroughly. Two novel techniques for segmenting off-line unconstrained cursive handwriting are being proposed and research papers are being produced, submitted and accepted in local and international conferences. Furthermore, one paper has also been submitted and accepted in the International journal of image processing. In addition to research, experiments are under way into skew removal, slant removal and line removal. Another research goal is to propose more preprocessing techniques in terms of feature extraction for recognizing segmented characters.

The following research papers were produced during the fellowship:

1/ Amjad Rehman and Dzulkifli Mohammad (2008). A Simple Segmentation Approach for Unconstrained Cursive Handwritten Words in Conjunction with the Neural Network. International Journal of Image Processing. Vol. 2, Issue 3, 29-35;

2/ Amjad Rehman and Dzulkifli Mohammad (2008). Off-line Cursive Handwriting Word Recognition: A Survey of Methods and Performances. Journal of Institute of Mathematics and Computer Science (Computer Science Series), Kolkata, India (Accepted);

3/ Amjad Rehman, Fajri Kurniawan and Dzulkifli Mohammad (2008). Off-line Cursive Handwriting Segmentation, A Heuristic Rule-based Approach. Journal of Institute of Mathematics and Computer Science (Computer Science Series) Kolkata, India. Vol. 19 (2), pp. 135-139;

4/ Dzulkifli Mohammad, Amjad Rehman Khan and Fajri Kurniawan (2008). A New Approach for Segmenting Difficult Cursive Handwritten Words from Benchmark Database. Proceedings of 4th International Conference on Information and Communication Technology and Systems (ICTS) Vol.1, pp.17-21; 5/ Dzulkifli Mohamad, Amjad Rehman (2008). Off-line Cursive Handwritten Word Segmentation, a New Approach. Proceedings of 4th Postgraduate Annual Research Seminar (PARS’ 08), UTM Skudai, Johor, Malaysia, pp.260-262.

 

January 2009