Abstract
Convolutional Neural Networks (CNNs) show state-of-the-art performance in handwritten word recognition. Existing CNNs operate on one language and their accuracy depends on dataset. Besides, since CNNs automatically extract features from the raw images, pre-processing is not usually applied. In this paper, we aim to achieve two goals: improving the accuracy of recognition task by pre-processing the raw images and designing a CNN structure for successful operation on similar languages (Persian and Arabic here). For doing this, a new CNN structure named PCNS is proposed as follows: images are firstly pre-processed to have the same pixel size. After that, images are given into GoogLeNet for high and low-level feature extraction. Finally, Support Vector Machine (SVM) makes the final classification. Experiments indicate that PCNS statistically outperforms other methods on Persian language (P value = 0.048) and provides competitive results with others on Arabic datasets. Test accuracies on Persian datasets are: Iranshahr (98.62%), Hoda (99.50%), and Farshid_LATP (98.83%). On Arabic datasets: MADBase (99.20%), HACDB (95.96%), and IFN/ENIT (97.65%).
Similar content being viewed by others
Data availability
Not applicable.
References
Akbarpour S (2011) Improved feature extraction and lexicon reduction methods classified by support vector machine for Farsi handwritten word recognition system. Dissertation, University Putra Malaysia
Alkhawaldeh RS (2021) Arabic (Indian) digit handwritten recognition using recurrent transfer deep architecture. Soft Comput 25(4):3131–3141. https://doi.org/10.1007/s00500-020-05368-8
Almodfer R, Xiong S, Mudhsh M, Duan P (2017) Multi-column deep neural network for offline Arabic handwriting recognition. Int Conf Artificial Neural Networks 260-267. https://doi.org/10.1007/978-3-319-68612-7_30
Amrouch M, Rabi M (2017) Deep neural networks features for Arabic handwriting recognition. Proc. Int. Conf. Adv Inform Technol, Services Syst 138-149. https://doi.org/10.1007/978-3-319-69137-4_14
Arani SA, Kabir E, Ebrahimpour R (Jan 2019) Handwritten Farsi word recognition using NN-based fusion of HMM classifiers with different types of features. Int J Image Graphics 19(01):1950001. https://doi.org/10.1142/S0219467819500013
Bagheri Noaparast K, Broumandnia A (2009) Persian handwritten word recognition using Zernike and Fourier–Mellin moments. 5th Int. Conf. Sci Electron Technol Inform Telecomm 1-7
Bayesteh E, Ahmadifard A, Khosravi H (2011) A lexicon reduction method based on clustering word images in offline Farsi handwritten word recognition systems. 7th IEEE Iranian Conf. On machine vision and image processing 1-5. https://doi.org/10.1109/IranianMVIP.2011.6121550
Bengio Y, Courville A, Vincent P (2013) Representation learning: a review and new perspectives. IEEE Trans Pattern Anal and Mach Intell 35(8):1798–1828. https://doi.org/10.1109/TPAMI.2013.50
Bonyani M, Jahangard S, Daneshmand M (2021) Persian handwritten digit, character and word recognition using deep learning. Int J Doc Anal Recogn (IJDAR) 24(1):133–143. https://doi.org/10.1007/s10032-021-00368-2
Cireşan D, Meier U, Schmidhuber J (2012) Multi-column deep neural networks for image classification. IEEE Conf on Comp Vision Patt Recogn:3642–3649. https://doi.org/10.1109/CVPR.2012.6248110
Dehghan M, Faez K, Ahmadi M, Shridhar M (2001) Handwritten Farsi (Arabic) word recognition: a holistic approach using discrete HMM. Pattern Recogn 34(5):1057–1065. https://doi.org/10.1016/S0031-3203(00)00051-0
Ebrahimpour R, Amini M, Sharifizadehi F (2011) Farsi handwritten recognition using combining neural networks based on stacked generalization. Int J Electrical Eng Informat 3(2):146–164
Ebrahimpour R, Amini M, Vahidi Shams A (2011) A new combination method based on different representation of data. Int J Hybrid Inf Technol 4(3):51–60
Ebrahimpour R, Sarhangi S, Sharifizadeh F (2011) Mixture of experts for Persian handwritten word recognition. Iranian J of Electr Electron Eng 7(4):217–224
Elleuch M, Tagougui N, Kherallah M (2015) Towards unsupervised learning for Arabic handwritten recognition using deep architectures. 22th Int. Conf. Neural Inform Process 363-372. https://doi.org/10.1007/978-3-319-26532-2_40
Elleuch M, Maalej R, Kherallah M (2016) A new design based-SVM of the CNN classifier architecture with dropout for offline Arabic handwritten recognition. Procedia Comput Sci 80:1712–1723. https://doi.org/10.1016/j.procs.2016.05.512
El-Sawy A, Hazem EB, Loey M (2016) CNN for handwritten Arabic digits recognition based on LeNet-5. Int. Conf. Advanc Intell Syst Inform 566-575. https://doi.org/10.1007/978-3-319-48308-5_54
Gao XW, Hui R (2016) A deep learning based approach to classification of CT brain images. SAI computing Conf. 28-31. https://doi.org/10.1109/SAI.2016.7555958
García S, Fernández A, Luengo J, Herrera F (2010) Advanced nonparametric tests for multiple comparisons in the design of experiments in computational intelligence and data mining: experimental analysis of power. Inf Sci 180(10):2044–2064. https://doi.org/10.1016/j.ins.2009.12.010
Haghighi F, Omranpour H (2021) Stacking ensemble model of deep learning and its application to Persian/Arabic handwritten digits recognition. Knowl-Based Syst 220:106940. https://doi.org/10.1016/j.knosys.2021.106940
Imani Z, Ahmadyfard A, Zohrevand A (2016) Holistic Farsi handwritten word recognition using gradient features. J AI Data Mining 4(1):19–25
Ioffe S, Szegedy C (2015) Batch normalization: accelerating deep network training by reducing internal covariate shift. Proc. 32nd Int. Conf. On Mach Learn 448–456.
Karpathy A, Toderici G, Shetty S, Leung T, Sukthankar R, Fei-Fei L (2014) Large-scale video classification with convolutional neural networks. Proc. of the 2014 IEEE Conf. On computer vision Patt Recogn 1725–1732. https://doi.org/10.1109/CVPR.2014.223
Krizhevsky A, Sutskever I, Hinton GE (2017) ImageNet classification with deep convolutional neural networks. Commun ACM 60(6):1097–1105. https://doi.org/10.1145/3065386
LeCun Y, Kavukcuoglu K, Farabet C (2010) Convolutional networks and applications in vision. Int Symp Circ Syst:253–256. https://doi.org/10.1109/ISCAS.2010.5537907
Loey M, El-Sawy A, EL-Bakry H (2017) Deep learning autoencoder approach for handwritten Arabic digits recognition. arXiv preprint arXiv:1706.06720.
Nanehkaran YA, Zhang D, Salimi S, Chen J, Tian Y, Al-Nabhan N (2020) Analysis and comparison of machine learning classifiers and deep neural networks techniques for recognition of Farsi handwritten digits. J Supercomput 77:1–30. https://doi.org/10.1007/s11227-020-03388-7
Nanehkaran YA, Zhang D, Salimi S, Chen J, Tian Y, Al-Nabhan N (2021) Analysis and comparison of machine learning classifiers and deep neural networks techniques for recognition of Farsi handwritten digits. J Supercomput 77(4):3193–3222. https://doi.org/10.1007/s11227-020-03388-7
Parseh M, Rahmanimanesh M, Keshavarzi P (2020) Persian handwritten digit recognition using combination of convolutional neural network and support vector machine methods. Int Arab J Inf Technol. 17(4):572–578. https://doi.org/10.34028/iajit/17/4/16
Poznanski A, Wolf L (2016) CNN-N-gram for handwriting word recognition. Proc. IEEE Conf. Comp Vision Patt Recogn:2305–2314. https://doi.org/10.1109/CVPR.2016.253
Sabzi R, Fotoohinya Z, Khalili A, Golzari S, Salkhorde Z, Behravesh S, Akbarpour S (2017) Recognizing Persian handwritten words using deep convolutional networks. IEEE Artif Intell Signal Process Conf:85–90. https://doi.org/10.1109/AISP.2017.8324114
Safarzadeh VM, Jafarzadeh P (2020) Offline Persian handwriting recognition with CNN and RNN-CTC. In 25th international computer conference, computer society of Iran (CSICC) 1-10, IEEE. https://doi.org/10.1109/CSICC49403.2020.9050073
Scherer D, Müller A, Behnke S (2010) Evaluation of Pooling Operations in Convolutional Architectures for Object Recognition. 20th Int. Conf. on Artificial Neural Networks 92–101. https://doi.org/10.1007/978-3-642-15825-4_10
Simard PY, Steinkraus D, Platt JC (2003) Best practices for convolutional neural networks applied to visual document analysis. Proc. of 7th Int. Conf. Docum Anal Recogn 958–962. https://doi.org/10.1109/ICDAR.2003.1227801
Sudholt S, Fink GA (2016) PHOCNet: a deep convolutional neural network for word spotting in handwritten documents. 15th Int. Conf. Front Handwriting Recogn 277-282. https://doi.org/10.1109/ICFHR.2016.0060
Szegedy C, Liu W, Jia W, Sermanet P, Reed S, Anguelov D, Erhan D, Vanhoucke V, Rabinovich A (2015) Going deeper with convolutions. Proc IEEE Conf Comp Vision Patt Recogn:1–9
Younessy Ghadikolaie MF, Kabir E, Razzazi F (2016) Sub-word based offline handwritten Farsi word recognition using recurrent neural network. ETRI J 38(4):703–713. https://doi.org/10.4218/etrij.16.0115.0542
Zamani Y, Souri Y, Rashidi H, Kasaei S (2015) Persian handwritten digit recognition by random Forest and convolutional neural networks. 9th Iranian IEEE Conf. Mach Vision Image Process 37-40. https://doi.org/10.1109/IranianMVIP.2015.7397499
Code availability
Not applicable.
Author information
Authors and Affiliations
Contributions
Not applicable.
Corresponding author
Ethics declarations
Conflict of interest
Not applicable.
Additional information
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Golzari, S., Khalili, A. & Sabzi, R. Combining convolutional neural networks with SVM classifier for recognizing Persian and Arabic handwritten words. Multimed Tools Appl 81, 33785–33799 (2022). https://doi.org/10.1007/s11042-022-13101-w
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11042-022-13101-w