Information Technology Reference
In-Depth Information
The proposed method for the handwritten character recognition using the descent
gradient approach of backpropagation algorithm showed the remarkable enhance-
ment in the performance. The use of binarization features along with backpropa-
gation feed forward neural network yielded the excellent recognition accuracy.
Although the success rate of 85.62 % is considered excellent but a scope of
improvement is always there. The performance of a recognition system mainly
depends on the quality of samples used for training and the techniques employed to
extract the features and the type of classi
er. Preprocessing techniques, feature
extraction techniques and the methodology used to select
the neural network
parameters can be improved to get further improved results.
Nevertheless, more work needs to be done especially on the test for more
complex handwritten characters. The proposed work can be carried out to recognize
English words of different character lengths after proper segmentation of the words
into isolated character images. In future, better pre-processing techniques will be
used. The skew and slat correction module will also be incorporated in the future
work. This module was not applied here in this experiment because it was assumed
that all the handwritten character samples are free from slant and skew. The
character images which are rotated at a certain angle will also be included in the
character recognition experiment carried out in future. Also, the various parameters
of the neural network architecture such as number of hidden neurons, number of
hidden layers, choice of activation function in hidden and output layers, learning
rate, momentum constant, cost function, training termination criterion etc. will also
be optimized by detailed rigorous investigation and experimentation.
References
Aburas, A. A., & Rehiel, S. A. (2008). New promising off-line tool for Arabic handwritten
character recognition based on JPEG2000 image compression. In Proceedings of the 3rd
International Conference on Introduction and Communication Technology
from Theory to
Alginahi, Y. (2010). Preprocessing techniques in character recognition. Character Recognition.In
Azar A. T. (2013). Fast neural network learning algorithms for medical applications. Neural
Azar, A. T., & El-Said, S. A. (2013). Probabilistic neural network for breast cancer classification.
Banashree, N. P., Andhre, D., Vasanta, R., & Satyanarayana, P. S. (2007). OCR for script
identification of Hindi (Devanagari) numerals using error diffusion Halftoning Algorithm with
neural classifier. International Journal of Computer, Information Science and Engineering, 1
(2), 281
—
285.
-
Bernsen, J.
(1986). Dynamic thresholding of grey-level
images.
In Proceedings of 8th
1255), Paris, France.
Bharath, A. & Madhvanath, S. (2008). FreePad: A novel handwriting-based text input for pen and
touch interfaces. In Proceedings of the 13th International Conference on Intelligent User
Interfaces (pp. 297
International Conference on Pattern Recognition (pp. 1251
-
300), New York, NY, USA. doi:
10.1145/1378773.1378814
.
-
Search WWH ::
Custom Search