Please use this identifier to cite or link to this item:
http://hdl.handle.net/10603/266023
Title: | Development and Analysis of Soft Computing Techniques in Performance for Recognition of Handwritten Curve scripts of Marathi Characters |
Researcher: | Sangvikar Jagdish Arunrao |
Guide(s): | Singh M P |
Keywords: | Science and Technology |
University: | Swami Ramanand Teerth Marathwada University |
Completed Date: | 17/01/2018 |
Abstract: | The research work presented over here deals with Development and analysis of soft computing techniques in performance for the recognition of handwritten curve scripts of Marathi Characters . This work includes the five neural network models and two proposed feature extraction methods. The five neural network models are RBF, Cascade, Elman, Feed-forward and Pattern Recognition neural networks. The two suggested feature extraction methods used are Edge detection and dilation and CLAHE feature extraction methods. newlineThe earlier experimental work focuses on the performance analysis of RBF, Cascade, Elman, Feed-forward and Pattern Recognition neural networks for handwritten character recognition of Marathi curve scripts. For this experiment, we have taken into account the six handwritten samples each of 48 Marathi characters formed by six different peoples. The sample set size here is 48*6 = 288. Out of 288, 192 samples (48*4 =192) are used for the experiment. For the sample set, the methodology followed here is.. Extract the features, create the network, train the network, simulate the network and plot the regression plots. newlineIn the subsequent part of the experiment, we have studied, analyzed and evaluated the performance of RBF, Cascade, Elman, Feed Forward and Pattern Recognition Networks using CLAHE method i.e Contrast-limited Adaptive Histogram Equalization method of feature extraction. The same sample set used here is comprised of all 288 characters. Methodology used here is same as above to carry out the experiments. Then we have studied and analyzed the performance of these five Neural Networks for character recognition. From the observation tables and comparative charts, it has been found that except Elman Network, the performance all of networks is improved with respect to CLAHE feature extraction method. The RBF neural networks, with regression value R=1 in both the experiment has shown a very good performance. newlineIn future work, it is planned to improve the performance of Elman network with the help of v |
Pagination: | 153p |
URI: | http://hdl.handle.net/10603/266023 |
Appears in Departments: | School of Computational Sciences |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
01_title.pdf | Attached File | 109.69 kB | Adobe PDF | View/Open |
02_certificate.pdf | 109.07 kB | Adobe PDF | View/Open | |
03_abstract.pdf | 79.32 kB | Adobe PDF | View/Open | |
04_declaration.pdf | 112.31 kB | Adobe PDF | View/Open | |
05_acknowledgements.pdf | 51.27 kB | Adobe PDF | View/Open | |
06_contents.pdf | 72.95 kB | Adobe PDF | View/Open | |
07_list_of_tables.pdf | 23.37 kB | Adobe PDF | View/Open | |
08_list_of_figures.pdf | 121.81 kB | Adobe PDF | View/Open | |
09_abbreviations.pdf | 19.34 kB | Adobe PDF | View/Open | |
10_chapter 1.pdf | 238.32 kB | Adobe PDF | View/Open | |
11_chapter 2.pdf | 517.89 kB | Adobe PDF | View/Open | |
12_chapter 3.pdf | 861.84 kB | Adobe PDF | View/Open | |
13_chapter 4.pdf | 616.78 kB | Adobe PDF | View/Open | |
14_chapter 5.pdf | 740.56 kB | Adobe PDF | View/Open | |
15_conclusion.pdf | 175.58 kB | Adobe PDF | View/Open | |
16_summary.pdf | 109.09 kB | Adobe PDF | View/Open | |
17_bibliography.pdf | 248.28 kB | Adobe PDF | View/Open |
Items in Shodhganga are licensed under Creative Commons Licence Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0).
Altmetric Badge: