ABSTRACT
Contained in the scope of bilingual script environment and considered to be one of the applications of evolutionary computation and intelligent systems, handwriting recognition is a complex task due to the existence of huge number of languages as the globe’s means of communication. In this attempt at handwriting recognition using computers, the bilingual system for Roman and Gurmukhi script accepts online handwritten text and converts the handwritten data to its equivalent digital form, and for which conversion, an important step after pre-processing is segmentation. With the need for different levels of segmentation to be performed to get the digital data from handwritten samples, the first step is to segment at the level of complete words. A word is complete by considering the connectivity of different strokes with one another and the space present between different stroke sets. Next, the segmentation is achieved at the level of unique strokes representing a single word, which is the stroke level segmentation helping to provide valuable information about the script of the word under consideration. So, script identification process directly depends on the results of segmentation phase at the level of strokes. After the identification of script of the word under consideration, the recognition engine of the respective script helps in the extraction of refined information about the presence of different characters in the segmented word. The benefits of recognizing handwriting obtained at different levels of segmentation are considerable.
