CHARACTER SEGMENTATION USING SPEED VALUE AND ACCELERATION ON RECOGNITION HANDWRITING TECHNIQUE

Abstract

Nowadays, handwriting recognition has been widely used for analyzing signature, reading address as well as
changing text to hypertext. Common steps in handwriting recognition system consist of pre-processing,
segmentation, feature extraction, classification and post-processing. This research describes segmentation
methods for handwriting character which acquired using online device. Characters used in this research are 19
latin alphabets. In the first step, pre-processing is taken place by removing noise manually. The next step is
calculating the speed of writing. The acquired speed value is further used to calculate acceleration by applying
low-passfilter. As shown in the figure, it is found that there are many peak change has been identified so that
low-pass filter is also used to reduce the number of segment. Too many segments may increase the difficulties of
the next process. This research employs 4 segmentation methods, such as Positive Negative Acceleration,
Positive Positive Acceleration, Negative Positive Acceleration and Minimum Speed. From the experiments, it is
identified that by applying minimum speed, character segmentation can be perform quite well

Keyword(s): Segmentation, Character, Speed, Acceleration, Low-pass Filter.

Subject Description: I.4 Image Processing and Computer Vision | I.4.6 Segmentation | Edge and feature
detection