Information Technology Reference
In-Depth Information
Fig. 1 Schematic diagram of the proposed character recognition system
Handwritten character images obtained by segmenting the handwritten words
were also involved for the proposed feature extraction and recognition experiment.
All the word images were preprocessed and the resultant thinned word images were
cut vertically into the isolated characters using the proposed vertical dissection
based segmentation approach.
4.2 Local Database Preparation
Character image samples written by 5 different people (age 15
50 years) were
collected where each writer contributed 5 samples of the complete English alphabet
(a
-
26 = 650) fresh character image samples were
gathered for the proposed experiment. 650 character images obtained by segmenting
the handwritten words are also collected for the proposed character recognition
experiment.
A database of the 1,300 handwritten character images has been prepared by
combining 650 character images obtained by the segmentation of the handwritten
word images and another 650 fresh handwritten character images acquired by
scanning the character samples contributed by 5 different persons each contributing
5 samples (a
z). In this way 650 (5
×
5
×
-
z). This database containing a total number of 1,300 character images
was used for the proposed character feature extraction and recognition experiment.
A single writer contributed 5 character samples and is shown in Fig.
2
.
-
Search WWH ::
Custom Search