Data capturing process for online Gurmukhi script recognition system

Online handwriting recognition systems require input of user´s handwriting with the help of devices like digitizers. The input generated by the digitizer is in the form of strokes and each stroke maintain the information about x and y coordinate values of each point appeared in its formation. To obtain the values of these points for each stroke in any implementation environment, some Ink objects are required. This paper presents a technique to capture handwritten data made by these strokes from the surface of digitizer using C#.Net implementation environment. The handwritten data is written by using the characters of Gurmukhi script. This script is used to write "Punjabi" language in India. In this paper object model used for Ink collection and recognition is also explained. The purpose of proposed technique is to collect Ink, find logical locations of the coordinate positions of different points of a stroke and identification of Ink strokes. The use of "InkCollector" and "InkOverlay" objects provides the real world Inking capability to the applications used for recognition of scripts. In this paper, during data capturing process 100% accuracy is achieved in collection of coordinate locations for each stroke and to find the total quantity of handwritten strokes by considering all 41 consonants, 12 vowels and 3 half characters of Gurmukhi script. Data capturing process is completed with the help and input from 30 different writers. The exact coordinate information fetched for each stroke further gives strength to the structural behavior of these strokes in recognition and classification phase.