Old Jawi manuscript: Digital recognition / Zaidi Razak

Zaidi, Razak (2016) Old Jawi manuscript: Digital recognition / Zaidi Razak. PhD thesis, University of Malaya.

Preview

PDF (Thesis PhD) Download (3449Kb) | Preview

Abstract

Jawi is an ancient writing style that originated from Arabic characters but has additional characters to suit certain local pronunciation in the Malay spoken language. Jawi has been the lingua franca in the Malay archipelagos long before the introduction of Roman characters in this part of the world. As such, there are thousands of old manuscripts on the socio-economic development, governance and history of the Malay Archipelagos written in Jawi. These old manuscripts are rich sources of knowledge and thus invaluable reference for learning the culture, traditions, and early history of Malaysia. However, these old manuscripts are not easily accessible for learning or research. Moreover, old Jawi manuscripts are not easily understood because the complexity of the writing, the overlapping of characters, physical condition of the manuscripts, and the absence of a consistent writing style. The mentioned problems have provided the motivation to conduct this research. Digitizing the old Jawi manuscripts and converting old Jawi to modern Jawi will provide solution to the problems. In this context, this research will examine Jawi writing style, its characters and the affinity between Jawi and Arabic characters. The literature review in this research looks at works conducted in the past on line and character segmentation as well as recognition systems. The information gathered is useful for the development of a line segmentation algorithm, a new character segmentation algorithm, as well as a recognition algorithm which is based on the use of a unique code and Hamming distance calculation. This research did not use artificial intelligence techniques to achieve the research goal. The algorithms developed in this research were evaluated against similar algorithms for other recognition methods. The evaluation results show that the proposed approach outperforms the algorithms of other recognition methods used for comparison. Future research in this domain could explore the implementation of a hardware-based Jawi recognition system, development of a character recognition technique to handle different character widths, designing comprehensive manuscript preservation system, and developing a tool for manuscript authentication.

Actions (For repository staff only : Login required)