Simplified Point to Point Correspondence of the Euclidean Distance for Online Handwriting Recognition
Petra Bilane, Éliane Youssef, B. Eter, Charles M Sarraf, Joseph Constantin · 2007
This paper describes an online handwriting recognition system. Our system presents a modified and less complex version of the point to point correspondence method that originally relies on a dynamic time warping algorithm. The system first passes through a training phase in which it is taught the handwriting of a certain person. The training consists of the person writing several times all the letters of the alphabet, data acquisition is done using a digitizer tablet. Training data for each character is then stored as pixels coordinates in the same order as their creation. In the recognition phase, the system recognizes a character written by the same person based on the previously done training. The obtained results were very encouraging; a recognition rate of 93.35 % for isolated lower case characters could be achieved relying only on the training done before the recognition phase without the need for a recognition database.