Improved word-level handwritten Indic script identification by integrating small convolutional neural networks

  • Soumya Ukil
  • Swarnendu Ghosh
  • Sk Md Obaidullah
  • K. C. Santosh
  • Kaushik Roy
  • Nibaran DasEmail author
Original Article


Handwritten document recognition has been an active domain of research in the field of computer vision for several years since 1914 with the development of handheld scanner for reading printed texts called “optophone”. In India, which has several different scripts in one document page, identifying them is a must to automate process: document understanding. We propose a novel technique in integrating convolutional neural networks (CNNs) for script identification. We combined small individually trainable small CNNs, and used several different levels of variation in the architectures of the individual CNNs. Such a collection of individually trainable modules vary with respect to the input image size, CNN’s depth and wavelet transformation. In our test, we used publicly available dataset of size 11K words (1K per script) from 11 different Indic Scripts: Bangla, Devanagari, Gujarati, Gurumukhi, Kannada, Malayalam, Oriya, Roman, Tamil, Telugu and Urdu. Several ensemble strategies were implemented such as max-voting and probabilistic voting are used in addition to other conventional approaches like feature concatenation. We achieved a maximum accuracy of 95.04%, and it outperforms the accuracy of the state-of-the-art techniques like AlexNet by 2.9% and more importantly, benchmark techniques as (for script identification) on the dataset by more than 4%.


Convolutional neural network Deep learning Haar wavelet transform Document analysis Indic script recognition More 



This work is supported by the project order no. SB/S3/EECE/054/2016, dated 25/11/2016, sponsored by SERB (Government of India) and carried out at the Centre for Microprocessor Application for Training Education and Research, CSE Department, Jadavpur University, Kolkata, India.


  1. 1.
    Ahmed SB, Naz S, Razzak MI, Rashid SF, Afzal MZ, Breuel TM (2016) Evaluation of cursive and non-cursive scripts using recurrent neural networks. Neural Comput Appl 27(3):603–613CrossRefGoogle Scholar
  2. 2.
    Anil R, Manjusha K, Kumar SS, Soman K (2015) Convolutional neural networks for the recognition of Malayalam characters. In: Proceedings of the 3rd international conference on frontiers of intelligent computing: theory and applications (FICTA) 2014. Springer, pp 493–500Google Scholar
  3. 3.
    Basu S, Das N, Sarkar R, Kundu M, Nasipuri M, Basu DK (2009) A hierarchical approach to recognition of handwritten bangla characters. Pattern Recognit 42(7):1467–1484zbMATHCrossRefGoogle Scholar
  4. 4.
    Bhattacharya U, Chaudhuri B (2005) Databases for research on recognition of handwritten characters of Indian scripts. In: Eighth international conference on document analysis and recognition, 2005. Proceedings. IEEE, pp 789–793Google Scholar
  5. 5.
    Bhattacharya U, Chaudhuri BB (2009) Handwritten numeral databases of Indian scripts and multistage recognition of mixed numerals. IEEE Trans Pattern Anal Mach Intell 31(3):444–457CrossRefGoogle Scholar
  6. 6.
    Bracewell RN, Bracewell RN (1986) The Fourier transform and its applications, vol 31999. McGraw-Hill, New YorkzbMATHGoogle Scholar
  7. 7.
    Brodić D, Amelio A, Milivojević ZN (2016) Language discrimination by texture analysis of the image corresponding to the text. Neural Comput Appl 29:1–22Google Scholar
  8. 8.
    Busch A, Boles WW, Sridharan S (2005) Texture for script identification. IEEE Trans Pattern Anal Mach Intell 27(11):1720–1732CrossRefGoogle Scholar
  9. 9.
    Das N, Sarkar R, Basu S, Saha PK, Kundu M, Nasipuri M (2015) Handwritten bangla character recognition using a soft computing paradigm embedded in two pass approach. Pattern Recognit 48(6):2054–2071CrossRefGoogle Scholar
  10. 10.
    Daubechies I (1990) The wavelet transform, time-frequency localization and signal analysis. IEEE Trans Inf Theory 36(5):961–1005MathSciNetzbMATHCrossRefGoogle Scholar
  11. 11.
    Dhanya D, Ramakrishnan A, Pati PB (2002) Script identification in printed bilingual documents. Sadhana 27(1):73–82zbMATHCrossRefGoogle Scholar
  12. 12.
    Garain U, Chakraborty M, Dasgupta D (2006) Recognition of handwritten indic script using clonal selection algorithm. In: Artificial immune systems, pp 256–266Google Scholar
  13. 13.
    Ghosh D, Dube T, Shivaprasad A (2010) Script recognition a review. IEEE Trans Pattern Anal Mach Intell 32(12):2142–2161CrossRefGoogle Scholar
  14. 14.
    Govindaraju V, Setlur S (2009) Guide to OCR for indic scripts. Springer, BerlinGoogle Scholar
  15. 15.
    Hangarge M, Santosh K, Pardeshi R (2013) Directional discrete cosine transform for handwritten script identification. In: 2013 12th International conference on document analysis and recognition (ICDAR). IEEE, pp 344–348Google Scholar
  16. 16.
    John J, Pramod K, Balakrishnan K (2012) Unconstrained handwritten Malayalam character recognition using wavelet transform and support vector machine classifier. Procedia Eng 30:598–605CrossRefGoogle Scholar
  17. 17.
    Kingma D, Ba J (2014) Adam: a method for stochastic optimization. arXiv preprint arXiv:1412.6980
  18. 18.
    Krizhevsky A, Sutskever I, Hinton G.E (2012) Imagenet classification with deep convolutional neural networks. In: Advances in neural information processing systems, pp 1097–1105Google Scholar
  19. 19.
    LeCun Y (1998) The mnist database of handwritten digits.
  20. 20.
    LeCun Y, Bottou L, Bengio Y, Haffner P (1998) Gradient-based learning applied to document recognition. Proc IEEE 86(11):2278–2324CrossRefGoogle Scholar
  21. 21.
    Mehrotra K, Jetley S, Deshmukh A, Belhe S (2013) Unconstrained handwritten devanagari character recognition using convolutional neural networks. In: Proceedings of the 4th international workshop on multilingual OCR. ACM, p 15Google Scholar
  22. 22.
    Neeba N, Jawahar C (2009) Empirical evaluation of character classification schemes. In: Seventh international conference on advances in pattern recognition, 2009. ICAPR’09. IEEE, pp 310–313Google Scholar
  23. 23.
    Obaidullah SM, Das N, Halder C, Roy K (2015) Indic script identification from handwritten document images an unconstrained block-level approach. In: 2015 IEEE 2nd international conference on recent trends in information systems (ReTIS). IEEE, pp 213–218Google Scholar
  24. 24.
    Obaidullah SM, Halder C, Santosh K, Das N, Roy K (2017) Phdindic\(\_\)11: page-level handwritten document image dataset of 11 official indic scripts for script identification. Multimedia Tools Appl 77:1–36Google Scholar
  25. 25.
    Obaidullah SM, Santosh K, Halder C, Das N, Roy K (2017) Automatic indic script identification from handwritten documents: page, block, line and word-level approach. Int J Mach Learn Cybern 10:1–20Google Scholar
  26. 26.
    Pal U, Chaudhuri B (2004) Indian script character recognition: a survey. Pattern Recognit 37(9):1887–1899CrossRefGoogle Scholar
  27. 27.
    Pal U, Jayadevan R, Sharma N (2012) Handwriting recognition in indian regional scripts: a survey of offline techniques. ACM Trans Asian Lang Inf Process (TALIP) 11(1):1CrossRefGoogle Scholar
  28. 28.
    Pal U, Sinha S, Chaudhuri B (2003) Multi-script line identification from indian documents. In: Seventh international conference on document analysis and recognition, 2003. Proceedings. IEEE, pp 880–884Google Scholar
  29. 29.
    Pati PB, Ramakrishnan A (2008) Word level multi-script identification. Pattern Recogn Lett 29(9):1218–1229CrossRefGoogle Scholar
  30. 30.
    Portnoff M (1980) Time-frequency representation of digital signals and systems based on short-time fourier analysis. IEEE Trans Acoust Speech Signal Process 28(1):55–69zbMATHCrossRefGoogle Scholar
  31. 31.
    Porwik P, Lisowska A (2004) The haar-wavelet transform in digital image processing: its status and achievements. Mach Graph Vis 13(1/2):79–98zbMATHGoogle Scholar
  32. 32.
    Rajput G, Anita H (2013) Handwritten script recognition at line level-a multiple feature based approach. Int J Eng Innovative Technol 3(4):90–95Google Scholar
  33. 33.
    Rani R, Dhir R, Lehal GS (2013) Script identification of pre-segmented multi-font characters and digits. In: 2013 12th International conference on document analysis and recognition (ICDAR). IEEE, pp 1150–1154Google Scholar
  34. 34.
    Roy K, Das S.K, Obaidullah SM (2011) Script identification from handwritten document. In: 2011 Third national conference on computer vision, pattern recognition, image processing and graphics (NCVPRIPG). IEEE, pp 66–69Google Scholar
  35. 35.
    Roy S, Das N, Kundu M, Nasipuri M (2017) Handwritten isolated bangla compound character recognition: a new benchmark using a novel deep learning approach. Pattern Recogn Lett 90:15–21CrossRefGoogle Scholar
  36. 36.
    Sarkhel R, Das N, Das A, Kundu M, Nasipuri M (2017) A multi-scale deep quad tree based feature extraction method for the recognition of isolated handwritten characters of popular indic scripts. Pattern Recognit 71:78–93CrossRefGoogle Scholar
  37. 37.
    Schenkel M, Guyon I, Henderson D (1995) On-line cursive script recognition using time-delay neural networks and hidden markov models. Mach Vis Appl 8(4):215–223CrossRefGoogle Scholar
  38. 38.
    Sharma MK, Dhaka VP (2016) Pixel plot and trace based segmentation method for bilingual handwritten scripts using feedforward neural network. Neural Comput Appl 27(7):1817–1829CrossRefGoogle Scholar
  39. 39.
    Singh PK, Mondal A, Bhowmik S, Sarkar R, Nasipuri M (2015) Word-level script identification from handwritten multi-script documents. In: Proceedings of the 3rd international conference on frontiers of intelligent computing: theory and applications (FICTA) 2014. Springer, pp 551–558Google Scholar
  40. 40.
    Singh PK, Sarkar R, Nasipuri M (2015) Offline script identification from multilingual indic-script documents: a state-of-the-art. Comput Sci Rev 15:1–28MathSciNetCrossRefGoogle Scholar
  41. 41.
    Singh PK, Sarkar R, Nasipuri M, Doermann D (2015) Word-level script identification for handwritten indic scripts. In: 2015 13th International conference on document analysis and recognition (ICDAR). IEEE, pp 1106–1110Google Scholar
  42. 42.
    Smith S (1997) Fourier transform properties. The scientist and engineers guide to digital signal processing. California Technical Publishing, San Diego, pp 185–208Google Scholar
  43. 43.
    Stanković RS, Falkowski BJ (2003) The haar wavelet transform: its status and achievements. Comput Electr Eng 29(1):25–44zbMATHCrossRefGoogle Scholar
  44. 44.
    Ubul K, Tursun G, Aysa A, Impedovo D, Pirlo G, Yibulayin T (2017) Script identification of multi-script documents: a survey. IEEE Access 5:6546–6559Google Scholar
  45. 45.
    Verma K, Sharma RK (2016) Comparison of HMM-and SVM-based stroke classifiers for Gurmukhi script. Neural Comput Appl 28:1–13CrossRefGoogle Scholar

Copyright information

© Springer-Verlag London Ltd., part of Springer Nature 2019

Authors and Affiliations

  1. 1.Jadavpur UniversityKolkataIndia
  2. 2.Aliah UniversityKolkataIndia
  3. 3.The University of South DakotaVermillionUSA
  4. 4.West Bengal State UniversityKolkataIndia

Personalised recommendations