Perekayasaan Histogram Orientasi Kecerunan
Mengesan Erotan dan Pencongan Manuskrip
Merong Mahawangsa


MOHD SANUSI BIN AZMI, KHAIRUDDIN OMAR & AZIZI ABDULLAH


ABSTRAK


Pengecaman Teks Optik (PTO) adalah popular dan telah diperkembangkan kepada teks jawi. Pengkajian ditumpukan kepada penentuan sudut erotan dan pencongan pada teks jawi. Terdapat pelbagai kaedah dalam menentukan sudut erotan dan pencongan teks jawi di antaranya Kuasa Dua Terkecil
(KDT) dan Histogram Orientasi Cerunan (HOC). Kaedah yang direkayasa ialah HOC dipelopori oleh Sun dan Si untuk mengesan erotan dan pencongan tulisan rumi. Teknik ini turut popular di kalangan penyelidik-penyelidik tulisan Jawi untuk mengesan erotan dan pencongan tulisan tangan Jawi. Walau bagaimanapun, kecekapan pengesanan sudut ini perlu diselidiki dan diperbaiki dengan algoritma rekayasa agar sesuai dengan kecekapan dan ketepatan sudut pengesanan. Perekayasaan algoritma HOC sedia ada mungkin dapat menyumbangkan kepada ketepatan dalam mengesan erotan dan pencongan tulisan Jawi. Pengujian dilaksanakan untuk membandingkan HOC asal dan yang direkayasa serta algoritma KDT.


ABSTRACT


Optic Text Recognition (PTO) is popular and has been extended to Jawi (Arabic character) text. The research is focused on the determination of the degree of skew and slant in Jawi text. There are a variety of methods to determine the degree of skew and slant in Jawi text like the Smallest Square
(KDT) and Gradient Orientation Histogram (HOC). The reengineered method is the HOC pioneered by Sun and Si to detect the skew and slant of Latin alphabet. This technique is also popular among Jawi researchers to detect the skew and slant of Jawi writing. However, the efficiency of the degree determination need to be investigated and enhanced with a reengineered algorithm so that it is suitable with the efficiency and accuracy of the determination degree. The reengineering of the present HOC algorithm may contribute to the accuracy in the skew and slant detection of Jawi writing. Tests are conducted to compare the original HOC and the enhanced HOC, as well as the KDT algorithm.

References