Keywords
Devanagari script
Image enhancement techniques
Convolutional neural networks (CNNs)
Fractional-order operators
How to Cite
Abstract
Handwritten character recognition is vital for document digitization and autonomous reading systems. This study focuses on enhancing handwritten Devanagari character recognition using deep learning models, specifically fine-tuned Convolutional Neural Networks (CNNs), combined with hybrid mathematical methods for image enhancement. Devanagari script, used for several South Asian languages, presents challenges due to its complexity and structural variations. To improve recognition accuracy, we propose a fuzzy-enabled Power-Law transformation for image enhancement, along with other techniques like Grunwald- Letnikov Fractional Differentiation (GLFD) and Atangana-Baleanu-Riemann (ABR). Experimental results show that the CNN model with Power-Law+ Fuzzy enhancement achieves the highest accuracy (98.02%), surpassing even MobileNetV2 (92.86%). The same method also yields impressive performance on the MNIST dataset (99.15% accuracy), demonstrating its effectiveness across different scripts. These findings highlight the benefits of integrating advanced preprocessing with deep learning for improved handwriting recognition, offering practical applications in multilingual document processing and OCR-based automation.
References
Acharya, S. and P. Gyawali, 2015 Devanagari handwritten character dataset. UCI Machine Learning Repository.
Agrawal, K. K. and M. S. Nair, 2019 Fractional order edge detection with GL fractional derivative mask. Signal, Image and Video Processing 13: 27–34.
Al-Ataby, A., W. Al-Nuaimy, and M. Al-Taee, 2021 Handwritten digit recognition using support vector machine and other traditional machine learning approaches. Engineering Technology Journal 39: 1131–1141.
Arora, S., L. Malik, S. Goyal, D. Bhattacharjee, M. Nasipuri, et al., 2024 Devanagari character recognition: A comprehensive literature review. IEEE Access.
Atangana, A. and D. Baleanu, 2016 New fractional derivatives with nonlocal and non-singular kernel: Theory and application to heat transfer model. Thermal Science 20: 763–769.
Das, S. and P. Gupta, 2013 Edge detection using fractional order differentiation and its applications. Procedia Computer Science 17: 301–308.
Deng, L., 2012 The MNIST database of handwritten digit images for machine learning research. IEEE Signal Processing Magazine 29: 141–142.
Garg, V. and K. Singh, 2012 An improved Grunwald–Letnikov fractional differential mask for image texture enhancement. International Journal of Advanced Computer Science and Applications 3: 130–135.
Gonzalez, R. C. and R. E. Woods, 2002 Digital Image Processing. Prentice Hall, second edition.
Hasan, M. M., M. A. Hossain, A. Y. Srizon, and A. Sayeed, 2023 Devanagari handwritten character recognition using fine-tuned deep convolutional neural network on trivial dataset. Multimedia Tools and Applications 82: 20671–20686.
Howard, A. G. et al., 2017 MobileNets: Efficient convolutional neural networks for mobile vision applications. arXiv preprint arXiv:1704.04861.
Jayadevan, R., S. R. Kolhe, P. M. Patil, and U. Pal, 2011 Offline recognition of Devanagari script: A survey. IEEE Transactions on Systems, Man, and Cybernetics, Part C (Applications and Reviews) 41: 782–796.
Khaparde, M. S. and V. H. Mankar, 2018 A comprehensive review on OCR for handwritten Devanagari script. International Journal of Engineering and Technology 7: 31–37.
Krizhevsky, A., I. Sutskever, and G. E. Hinton, 2017 ImageNet classification with deep convolutional neural networks. Communications of the ACM 60: 84–90.
Kumar, R., A. Gaur, K. Bhushan, and M. Singh, 2021 Knowledge extraction in digit recognition using MNIST dataset: Evolution in handwriting analysis. International Journal of Applied Engineering Research 16: 845–852.
LeCun, Y., L. Bottou, Y. Bengio, and P. Haffner, 1998 Gradient-based learning applied to document recognition. Proceedings of the IEEE 86: 2278–2324.
LeCun, Y., C. Cortes, and C. J. C. Burges, 2010 MNIST handwritten digit database. Available at http://yann.lecun.com/exdb/mnist.
Liu, C. L., K. Nakashima, H. Sako, and H. Fujisawa, 2003 Handwritten digit recognition: Benchmarking of state-of-the-art techniques. Pattern Recognition 36: 2271–2285.
Naidu, G., T. Zuva, and E. M. Sibanda, 2023 A review of evaluation metrics in machine learning algorithms. In Artificial Intelligence Application in Networks and Systems, edited by R. Silhavy and P. Silhavy, Springer, Cham.
Narang, S. R., M. Kumar, and M. K. Jindal, 2021 DeepNet Devanagari: A deep learning model for Devanagari ancient character recognition. Multimedia Tools and Applications 80: 20671–20686.
Pal, U. and B. B. Chaudhuri, 2004 Indian script character recognition: A survey. Pattern Recognition 37: 1887–1899.
Podlubny, I., 1999 Fractional Differential Equations. Academic Press.
Ramesh Babu, N., A. S. Joshua, P. Balasubramaniam, and A. Tiwari, 2024 Fuzzy-driven image enhancement via ABR-fractal-fractional differentiation. Information Sciences 675: 120741.
Sandler, M., A. Howard, M. Zhu, A. Zhmoginov, and L. C. Chen, 2018 MobileNetV2: Inverted residuals and linear bottlenecks. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.
Sarangi, P. K., S. Singh, and C. Singla, 2020 Feature selection for character recognition of handwritten Devanagari and Odia scripts. International Journal of Engineering Research and Technology 13: 1974–1982.
Sethi, N. and A. Singh, 2020 Handwritten Devanagari character recognition using deep convolutional neural network. Procedia Computer Science 173: 215–222.
Sharma, S. and D. Kumar, 2021 A fuzzy-based ABR image enhancement technique for improved handwritten character recognition. Journal of Ambient Intelligence and Humanized Computing 12: 10457–10469.
Yacouby and Axman, 2020 Probabilistic extension of precision, recall, and F1 score for more thorough evaluation of classification models. In Proceedings of Eval4NLP.
Zhao, H. H. and H. Liu, 2020 Multiple classifiers fusion and CNN feature extraction for handwritten digits recognition. Granular Computing 5: 411–418.
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.