JAVANESE CHARACTERS IMAGE SEGMENTATION FROM DOCUMENT IMAGE OF HAMONG TANI

Abstract

Script image segmentation of a document image is the most decisive step to the success of the process of transliteration of the script image into another script, such as automatically transliterating a printed Javanese manuscript image into a Latin manuscript. This paper gives an example of the application of profile projection modification to the segmentation of Javanese script document image of the entire 87 pages of the document image of HamongTani book. Based on the output of the developed system, the average percentage of correctness is 84.255% with the average standard deviation of 14.093%. This value of average percentage of correctness shows that the model developed for the Java script document image segmentation of the HamongTani book is relatively good.