Structuring Elements of Hit or Miss to Identify Pattern of Benchmark Latin Alphabets Strokes

Norzehan Sakamat (Universiti Teknologi MARA)
Noor Elaiza Abd Khalid (Universiti Teknologi MARA)
Inda Ishadah Nazrul Azha (Universiti Teknologi MARA)

Indonesian Journal of Electrical Engineering and Computer Science
Vol 12, No 1: October 2018

Identification of features in correct alphabet stroke formation is a primary factor in acknowledging handwriting legibility. Hit–or-Miss transform is a morphology operator that is often applied to identify geometric features. Identifying the correct structuring elements (SEs) provides a good geometric feature extraction. Therefore this research proposes to identify generic SEs representing alphabet strokes. The handwriting font used is Syazalina83v3 which is  popularly use in teaching writing for lower primary school children in Malaysia. The methodology consists of four phases which are alphabet selection; image pre-processing; manual measurement and hit or miss algorithm with single and various combination of 2x2, 3x3 and 5x5 SE window size; and SEs performance using Pearson correlation.  The  combination of horizontal and vertical, right diagonal and left diagonal SEs performs well with very strong correlation in detecting  Simple Straight Line(SSL), Complex Straight Line(CSL) , Curve Line (CL) and Combination of Curve Line and Simple Straight Line(CLSSL) based on Evan’s correlation guide.