Fuzzy Base Artificial Neural Network Model for Text Extraction from Images

Journal of Critical Reviews ISSN- 2394-5125 Vol 7, Issue 6, 2020

Review Article

FUZZY BASE ARTIFICIAL NEURAL NETWORK MODEL FOR TEXT EXTRACTION FROM IMAGES

V. Lakshman Narayana1, B. Naga Sudheer2, Venkata Rao Maddumala3,P.Anusha4

1Department of Information Technology, Vignan’s Nirula Institute of Technology & Science for Women, Peda Palakaluru, Guntur, Andhra Pradesh, India. [email protected] 2Asst.Professor, Department of IT, Vignan's Foundation for Science, Technology & Research (Deemed to be University), Vadlamudi, Guntur, Andhra Pradesh, India. [email protected] 3Department of Information Technology, Vignan’s Nirula Institute of Technology & Science for Women, Peda Palakaluru, Guntur, Andhra Pradesh, India. 4Department of Information Technology, Vignan’s Nirula Institute of Technology & Science for Women, Peda Palakaluru, Guntur, Andhra Pradesh, India.

Received: 15.02.2020 Revised: 19.03.2020 Accepted: 05.04.2020

Abstract Content Extraction assumes a significant job in discovering essential and important data. Content extraction includes discovery, restriction, following, binarization, extraction, improvement and acknowledgment of the content from the given picture. This paper proposes a bi-leveled picture characterization framework to group printed and transcribed reports into totally unrelated predefined classes. In the present article, we propose a Fuzzy guideline guided novel strategy that is utilitarian without any outer intercession during execution. The Fuzzy validation processor utilizes subtleties of divider speed and force, just as change, to decide if the deliberate reverberation signal part to be separated genuinely speaks to the divider speed as it were. Test results propose that this methodology is a proficient one in contrast with various different strategies widely tended to in writing. At long last, the exhibition of the proposed framework is contrasted and the current frameworks and it is seen that proposed framework performs superior to numerous different frameworks.

Key words: Fuzzy Rule, Image Extraction, Fuzzy Image Processing, Text Extraction, Fuzzy Inference System, Artificial Neural Network

INTRODUCTION In customary figuring approach, the prime contemplations are RELATED WORK accuracy, sureness, and meticulousness [1],[2],[3]. Conversely, The creators in proposed a changed strategy utilizing Fuzzy the main rules of delicate registering spin around the validation which is rule based [26][27][28]. In a Fuzzy based accompanying: resistance for imprecision, vulnerability, strategy was intended for recognizing edge without setting the halfway truth and guess. Here Fuzzy validation is utilized to edge esteem. In the edge of the dim scale picture is resolved recognize the edges of pictures [4],[5],[6]. Edge recognition utilizing Fuzzy validation [29][30][31]. The methodology alludes to the game-plan of finding pointed inconsistencies in a dependent on irregularity, will in general segment a picture by picture. Delicate registering is a best in class promising field distinguishing segregated focuses, lines and edges as indicated that incorporates Fuzzy validation, hereditary designing, by unexpected changes in dark levels in two adjoining districts developmental calculation and neural systems [7],[8],[9]. in the scene[32][33][34]. The issues of picture division These days, numerous other picture mining frameworks have become progressively questionable and extreme with regards likewise appeared which concentrate and procedure content to managing uproarious pictures [35][36][37]. characters, words, and lines from the heterogeneous arrangement of multi-text style, multi-size, multi- situated[24],[25], multi-hued, multi-lingual and multi-content records[10],[11],[12]. The fields of printed character acknowledgment and content separation for non-Indic, for example, Latin, Chinese, Japanese and Korean contents are now experienced [13], [14]. A picture or an example can be perceived utilizing earlier information or the factual data removed from the picture or the example [23]. Expecting that division and standardization shelter been done, we center on the subtask of item acknowledgment and check, and show the exhibition utilizing a few arrangements of pictures [15],[16]. These pictures present many testing research issues in content extraction and acknowledgment [17],[18],[19]. Content extraction from pictures have numerous valuable applications in report investigation , location of vehicle tag, examination of article with tables, maps, outlines, charts and so forth., Finally, in the last advance, a hunt procedure perceives the data from Fig. 1: Text Extraction Process the structure[22]. Presently, if another example is experienced, the machine distinguishes the structure where A Fuzzy based edge location channel that goes through two the info design has a place, and dependent on the structure the phase of procedure to expel clamor from grayscale pictures example is arranged [20],[21]. are utilized in pictures can be comprehensively ordered into Document pictures, Caption content pictures and Scene content pictures[38][39][40]. Extraction of content in archives

Journal of critical reviews 350

FUZZY BASE ARTIFICIAL NEURAL NETWORK MODEL FOR TEXT EXTRACTION FROM IMAGES with content on complex shading foundation is troublesome because of multifaceted nature of the foundation and stir up of Procedure color(s) of fore-ground content with shades of 1. Fuzzy all info esteems into Fuzzy participation capacities. foundation[41][42]. 2. Execute every single appropriate principle in the standard base to figure the Fuzzy yield capacities. The superimposed content is an amazing wellspring of 3. De-fuzzy the Fuzzy yield capacities to get "fresh" yield significant level semantics. These content events could be esteems. identified, portioned, and perceived naturally for ordering, recovery and outline. .In proposed edge discovery utilizing Fuzzy validation in tangle lab [43][44]. This work depends on sixteen guidelines, which separate the objective pixel. The proposed edge discovery techniques without setting limit esteem [45][46]. It uses the ostensible 2 × 2 visor that skim over the whole picture pixel by pixel. Fuzzy induction framework and conventional edge administrators are consolidated in. The paper classified different existing division systems into three classes: 1. Trademark highlights thresholding or grouping 2. Edge location and 3. Locale extraction. The division systems were outlined and remarks were given on the upsides and downsides of each approach. The edge determination plans dependent on dim level histogram and Fig. 2: Methodology block diagram neighborhood properties just as dependent on auxiliary, textural and syntactic systems were portrayed[47][48]. Progressions of tasks are performed on the information picture (In testing just as preparing stage) during the pre- Fuzzy frameworks and Artificial Neural Network (ANN) are handling. It helps in upgrading the picture rendering and delicate registering ways to deal with displaying master makes the picture appropriate for division. So as to accomplish conduct. The objective is to impersonate the activities of a these objectives: commotion sifting, transformation to twofold specialist who tackles complex issues. As it was, rather than and smoothing activities are performed on the info picture. researching the issue in detail, one sees how a specialist Figure shows a case of picture standardization. effectively handles the issue and gets information by guidance or potentially learning. So as to extricate a shape from a picture, it is important to distinguish it from the foundation components [49]. This should be possible by thinking about the force data or by looking at the pixels against a given layout Fig. 3: Tilt input image [51] [52]. Content extraction includes location, limitation, following, binarization, extraction, upgrade and acknowledgment of the content from the given picture. A few strategies have been produced for separating the content from a picture. The techniques referred to right now message Fig. 4: Image after processing is performed extraction in pictures are grouped by various kinds of pictures [50]. An enormous number of approaches have been proposed for content extraction from pictures. The current work on content METHODOLOGY extraction from pictures can be grouped by various criteria. It is fascinating that all techniques perpetually performed The pixels are extracted from the image using the following ineffectively for in any event a couple of occurrences. equation. Accordingly it was seen that any single calculation couldn't be effective for all uproarious picture types, even in a solitary application space. Associated parts are shaped considering the stroke width of two neighboring pixels. Letter up-and-comers are then assembled which gives the identified content the overall design of the proposed framework. Fuzzy sets will be This article arranges as indicated by the various sorts of sets with limits which are not exact. The information pixels are picture, examines those calculations and talks about the isolated into Fuzzy sets expressly high contrast though the presentation assessment. yield pixel is isolated into three Fuzzy sets explicitly dark, white and edge. Fuzzy picture handling is the assortment of all methodologies that comprehend, speak to and process the pictures, their fragments and highlights as Fuzzy sets. Regular Fuzzy validation has that ability of exploiting the methodologies of example grouping include bunching administrators' understanding and the quick information preparing tests and partner bunches to given classes. This preparing capacity of PCs. Fuzzy enrollment capacities are issue turns out to be progressively recalcitrant when the characterized for each term set of each semantic variable in quantity of highlights utilized for characterization increments. the principles. The human propelled highlights of this Increasingly unpredictable pictures can be disintegrated into a decreased standard set are then fused in a numerous neural structure of straight forward shapes particular methodology system combination approach. The Fuzzy vital is used in the parcels the order task into some sub characterization errands, combination of the systems prepared with various capabilities. fathom each sub-grouping task, and in the end incorporates The extracted pixels are sub divided into clusters for accurate the outcomes to acquire the last arrangement result. As it extraction as were, apportioning of the order task is completed with the end goal that each sub issue can be settled in a module by abusing the neighborhood vulnerabilities and misusing the worldwide vulnerabilities can consolidate the consequences of the considerable number of modules. Form the clusters created, each character is recognized as PL (di) = PL (1) + 10 α log10 (di) + 10 log10 (γ, di)

Journal of critical reviews 351

FUZZY BASE ARTIFICIAL NEURAL NETWORK MODEL FOR TEXT EXTRACTION FROM IMAGES

The characters are grouped and recognized as words using the following equation. The process for text extraction is accurate by character recognition process.

The process of text extraction from image is depicted in Figure 5.

Fig. 7: Accuracy in detection levels

CONCLUSION It gives us a general technique to consolidate estimated numerical data into a typical system a joined Fuzzy standard base that hypothetically engages both numerical and semantic data. There is a great deal of opportunity in picking the Fig. 5: Text Extraction from Images enrollment areas in the said structure. Indeed, this happens to be one of the key difficulties. This paper gives a wide RESULT AND ANALYSIS investigation of the different content extraction systems and The Handwritten Character Recognition framework was tried calculations proposed before. This paper additionally on a few distinctive checked manually written pictures is uncovered a presentation correlation table of various proposed with various styles. The outcomes are profoundly procedures that was proposed before for content extraction reassuring. All pictures of various classifications can be from a picture. Each approaches its own advantages and recognized by means of their homogeneousness or highlight limitations. A correlation with related work has been qualities. All the thresholding strategies are commonly introduced. ANNs have been prepared for this reason with founded on the qualities of one or a few highlights, which will different kinds of information tests and that is the reason the assist us with building a versatile component guided by some created program has a capacity to test and order the previously, settled techniques. On account of the test pictures information character into 52 distinct classes with a precision to the best outcomes we just utilized the first pictures and the of over 95%. changed variants with erosive clamor 60% and 100%.The actuality of utilizing two Fuzzy likelihood appropriations and REFERENCES two collection administrators permits four mixes. We are 1. Tang X, Gao X, Liu J, Zhang H(2002). "A Spatial-Temporal utilizing diverse clamor levels and utilize all the current Approach for Video Caption Detection And Recognition", procedures and our proposed strategy to separate pictures. IEEE Transactions on Neural Networks, Vol. 13, No. 4. The table 1 depicts the detection values. 2. Hazem M. El-Bakry, “A New High Speed Neural Model for Character Recognition Using Cross Correlation and Table 1: Detection rate Values Matrix Decomposition,” International Journal of Signal Fuzzy Based ANN CRM Processing, vol.2, no.3, 2005, pp. 183-202. Method 3. Lakshman Narayana Vejendla and A Peda Gopi, (2019),” Avoiding Interoperability and Delay in Healthcare Dataset Number of Sample Number of Sample Monitoring System Using Block Chain Technology”, Recognitio s Recognitio s Revue d'Intelligence Artificielle , Vol. 33, No. 1, n Rate n Rate 2019,pp.45-48. (%) (%) 4. Gopi, A.P., Jyothi, R.N.S., Narayana, V.L. et al. (2020), Training 36,854 76 56,234 93 “Classification of tweets data based on polarity using Validatio 45,875 70 63,851 88 improved RBF kernel of SVM”. Int. j. inf. tecnol. (2020). n https://doi.org/10.1007/s41870-019-00409-4. Test 56,458 68 75,845 92 5. A Peda Gopi and Lakshman Narayana Vejendla, (2019),” Certified Node Frequency in Social Network Using Parallel Diffusion Methods”, Ingénierie des Systèmes d' Information, Vol. 24, No. 1, 2019, pp.113-117. 6. Lakshman Narayana Vejendla and Bharathi CR, (2018). “Multi-mode Routing Algorithm with Cryptographic Techniques and Reduction of Packet Drop using 2ACK scheme in MANETs”, Smart Intelligent Computing and Applications, Vo1.1, pp.649-658. 7. Lakshman Narayana Vejendla and Bharathi C R, (2018), “Effective multi-mode routing mechanism with master- slave technique and reduction of packet droppings using 2-ACK scheme in MANETS”, Modelling, Measurement and Control A, Vol.91, Issue.2, pp.73-76. 8. Lakshman Narayana Vejendla, A Peda Gopi and N.Ashok Kumar, (2018). “Different techniques for hiding the text information using text steganography techniques: A survey”, Ingénierie des Systèmes d'Information, Vol.23, Issue.6, pp. 115-125. Fig. 6: Character Extraction Process 9. A Peda Gopi and Lakshman Narayana Vejendla (2018), “Dynamic load balancing for client server assignment in

Journal of critical reviews 352

FUZZY BASE ARTIFICIAL NEURAL NETWORK MODEL FOR TEXT EXTRACTION FROM IMAGES

distributed system using genetic algorithm”, Ingénierie Qualitative Analysis”, J. of Computation in Biosciences des Systèmes d'Information, Vol.23, Issue.6, pp. 87-98. and Engineering, Vol. 2, No.3, pp.1-4. 10. Lakshman Narayana Vejendla and Bharathi C 27. RSM Lakshmi Patibandla, Santhi Sri Kurra and H.-J. R,(2017),“Using customized Active Resource Routing and Kim,(2014), “Electronic resource management using Tenable Association using Licentious Method Algorithm cloud computing for libraries”, International Journal of for secured mobile ad hoc network Management”, Applied Engineering Research, Vol.9,pp. 18141-18147. Advances in Modeling and Analysis B, Vol.60, Issue.1, 28. Ms.R.S.M. Lakshmi Patibandla Dr. Ande Prasad and pp.270-282. Mr.Y.R.P. Shankar,(2013), “SECURE ZONE IN CLOUD”, 11. Lakshman Narayana Vejendla and Bharathi CR, (2017), International Journal of Advances in Computer Networks “Identity Based Cryptography for Mobile ad hoc and its Security, Vol.3, No.2, pp.153-157. Networks”, Journal of Theoretical and Applied 29. Patibandla, R.S.M. Lakshmi et al., (2016). “Significance of Information Technology, Vol.95, Issue.5, pp.1173-1181. Embedded Systems to IoT.” International Journal of EID: 2-s2.0-85015373447 Computer Science and Business Informatics, Vol.16, No.2, 12. Lakshman Narayana Vejendla and A Peda Gopi, (2017),” pp.15-23. Visual cryptography for gray scale images with enhanced 30. Anveshini Dumala and S. Pallam Setty. (2020),“LANMAR security mechanisms”, Traitement du Signal, Vol.35, routing protocol to support real-time communications in No.3-4, pp.197-208. MANETs using Soft computing technique”, 3rd 13. A Peda Gopi and Lakshman Narayana Vejendla, (2017),” International Conference on Data Engineering and Protected strength approach for image steganography”, Communication Technology (ICDECT-2019), Springer, Traitement du Signal, Vol.35, No.3-4, pp.175-181. Vol. 1079, pp. 231-243. 14. Lakshman Narayana Vejendla and A Peda Gopi, (2020),” 31. Anveshini Dumala and S. Pallam Setty. Design and Analysis of CMOS LNA with Extended (2019),“Investigating the Impact of Network Size on Bandwidth For RF Applications”, Journal of Xi'an LANMAR Routing Protocol in a Multi-Hop Ad hoc University of Architecture & Technology, Vol. 12, Issue. 3, Network”, i-manager’s Journal on Wireless pp.3759-3765. Communication Networks (JWCN), Volume 7, No. 4, 15. Chaitanya, K., and S. Venkateswarlu, (2016),"Detection of pp.19-26. Blackhole & Greyhole Attacks in Manets based on 32. Anveshini Dumala and S. Pallam Setty. Acknowledgement based Approach." Journal of (2019),“Performance analysis of LANMAR routing Theoretical and Applied Information Technology 89.1: protocol in SANET and MANET”, International Journal of 228. Computer Science and Engineering (IJCSE) – Vol. 7,No. 5, 16. Patibandla R.S.M.L., Kurra S.S., Mundukur N.B. (2012), pp.1237-1242. “A Study on Scalability of Services and Privacy Issues 33. Anveshini Dumala and S. Pallam Setty. (2018), “A in Cloud Computing”. In: Ramanujam R., Ramaswamy S. Comparative Study of Various Mobility Speeds of Nodes (eds) Distributed Computing and Internet Technology. on the Performance of LANMAR in Mobile Ad hoc ICDCIT 2012. Lecture Notes in Computer Science, vol Network”, International Journal of Computer Science and 7154. Springer, Berlin, Heidelberg Engineering (IJCSE) – Vol. 6, No. 9, pp. 192-198. 17. Patibandla R.S.M.L., Veeranjaneyulu N. (2018), “Survey 34. Anveshini Dumala and S. Pallam Setty. on Clustering Algorithms for Unstructured Data”. In: (2018),“Investigating the Impact of IEEE 802.11 Power Bhateja V., Coello Coello C., Satapathy S., Pattnaik P. Saving Mode on the Performance of LANMAR Routing (eds) Intelligent Engineering Informatics. Advances in Protocol in MANETs”, International Journal of Scientific Intelligent Systems and Computing, vol 695. Springer, Research in Computer Science and Management Studies Singapore (IJSRCSMS) – Vol.7, No. 4. 18. Patibandla, R.S.M.L., Veeranjaneyulu, N. (2018), 35. Anveshini Dumala and S. Pallam Setty. (2016),“Analyzing “Performance Analysis of Partition and Evolutionary the steady state behavior of RIP and OSPF routing Clustering Methods on Various Cluster Validation protocols in the context of link failure and link recovery Criteria”, Arab J Sci Eng, Vol.43, pp. 4379–4390. in Wide Area Network”, International Journal of 19. RSM Lakshmi Patibandla, Santhi Sri Kurra and N. Computer Science Organization Trends (IJCOT) – Vol. 34 Veeranjaneyulu, (2015), “A Study on Real-Time Business No 2, pp.19-22. Intelligence and Big Data”, Information Engineering, 36. Anveshini Dumala and S. Pallam Setty. Vol.4,pp.1-6. (2016),“Investigating the Impact of Simulation Time on 20. K. Santhisri and P.R.S.M. Lakshmi,(2015), “ Comparative Convergence Activity & Duration of EIGRP, OSPF Routing Study on Various Security Algorithms in Cloud Protocols under Link Failure and Link Recovery in WAN Computing”, Recent Trends in Programming Languages, Using OPNET Modeler”, International Journal of Vol.2, No.1, pp.1-6. Computer Science Trends and Technology (IJCST) – Vol. 4 21. K. Santhi Sri and PRSM Lakshmi, (2017), “DDoS Attacks, No. 5, pp. 38-42. Detection Parameters and Mitigation in Cloud 37. Vellalacheruvu Pavani and I. Ramesh Babu (2019) Environment”, IJMTST, Vol.3, No.1, pp.79-82. ,”Three Level Cloud Storage Scheme for Providing Privacy 22. P.R.S.M. Lakshmi, K. Santhi Sri and Dr.N. Veeranjaneyulu, Preserving using Edge Computing”, International Journal (2017), “A Study on Deployment of Web Applications of Advanced Science and Technology Vol. 28, No. 16, pp. Require Strong Consistency using Multiple Clouds”, 1929 – 1940. IJMTST, Vol.3, No.1, pp.14-17. 38. VellalacheruvuPavani and I. Ramesh Babu,”A Novel 23. P.R.S.M. Lakshmi, K. Santhi Sri and M.V. Bhujanga Ra0, Method to Optimize the Computation Overhead in Cloud (2017), “Workload Management through Load Balancing Computing by Using Linear Programming” ,International Algorithm in Scalable Cloud”, IJASTEMS, Vol.3, No.1, Journal of Research and Analytical Reviews May 2019, pp.239-242. Volume 6, Issue 2,PP.820-830.. 24. K. Santhi Sri, P.R.S.M. Lakshmi, and M.V. Bhujanga Ra0, 39. Anusha Papasani and Nagaraju Devarakonda, (2016), (2017), “A Study of Security and Privacy Attacks in Cloud "Improvement of Aomdv Routing Protocol in Manet and Computing Environment”, IJASTEMS, Vol.3, No.1, pp. 235- Performance Analysis of Security Attacks", International 238. Journal of Research in Computer Science & Engineering , 25. RSM Lakshmi Patibandla and N. Veeranjaneyulu, (2018), Vol.6, No.5, pp. 4674-4685. “Explanatory & Complex Analysis of Structured Data to 40. Sk. Reshmi Khadherbhi, K. Suresh Babu , Big Data Search Enrich Data in Analytical Appliance”, International Space Reduction Based On User Perspective Using Map Journal for Modern Trends in Science and Technology, Reduce ,International Journal of Advanced Technology Vol. 04, Special Issue 01, pp. 147-151. and Innovative Research Volume.07, IssueNo.18, 26. RSM Lakshmi Patibandla, Santhi Sri Kurra, Ande Prasad December-2015, Pages: 3642-3647 and N. Veeranjaneyulu, (2015), “Unstructured Data:

Journal of critical reviews 353

FUZZY BASE ARTIFICIAL NEURAL NETWORK MODEL FOR TEXT EXTRACTION FROM IMAGES

41. B.V. Suresh kumar, Sk. Reshmi Khadherbhi ,BIG-IOT Framework Applications and Challenges: A Survey Volume 7, Issue VII, JULY/2018 pg.no 1257-1264 42. P. Sandhya Krishna, Sk. Reshmi Khadherbhi, V. Pavani, Unsupervised or Supervised Feature Finding For Study of Products Sentiment ,International Journal of Advanced Science and Technology, Vol 28 No 16 (2019). 43. K. Santhi Sri, Dr. Ande Prasad (2013), “A Review of Cloud Computing and Security Issues at Different Levels in Cloud Computing”, International Journal on Advanced Computer Theory and Engineering Vol. 2, pp 67-73. 44. K. Santhi Sri, N. Veeranjaneyulu (2018), “A Novel Key Management Using Elliptic and Diffie-Hellman for Managing users in Cloud Environment”, Advances in Modelling and Analysis B, Vol.61, No.2, pp 106-112. 45. K. Santhi Sri, N. Veeranjaneyulu (2019), “Decentralized Key Management Using Alternating Multilinear Forms for Cloud Data Sharing with Dynamic Multiprivileged Groups”, Mathematical Modelling of Engineering Problems,Vol.6,No.4,pp511-518. 46. S. Sasikala, P.Sudhakar, “interpolation of CFA color Images with Hybrid image denoising”, 2014 Sixth International Conference on Computational Intelligence and Communication Networks, DOI 10.1109/.53 193 DOI 10.1109/CICN.2014.53, pp. 193-197. 47. Me. Jakeera Begum and M. Venkata Rao, (2015). “Collaborative Tagging Using CAPTCHA” International Journal of Innovative Technology and Research, Volume No.3, Issue No.5, pp, 2436 – 2439. 48. L.Jagajeevan Rao, M. Venkata Rao, T.Vijaya Saradhi (2016). “How The Smartcard Makes the Certification Verification Easy” Journal of Theoretical and Applied Information Technology, Vol.83. No.2, pp. 180-186. 49. Venkata Rao Maddumala, R. Arunkumar, and S. Arivalagan (2018)“An Empirical Review on Data Feature Selection and Big Data Clustering” Asian Journal of Computer Science and Technology Vol.7 No.S1, pp. 96- 100. 50. Singamaneni Kranthi Kumar, Pallela Dileep Kumar Reddy, Gajula Ramesh, Venkata Rao Maddumala, (2019), “Image Transformation Technique Using Steganography Methods Using LWT Technique”, Traitement du Signalvol 36, No 3, pp. 233-237. 51. Chaitanya, K., and S. Venkateswarlu, (2016),"Detection of Blackhole & Greyhole Attacks in Manets Based on Acknowledgement Based Approach." Journal of Theoretical and Applied Information Technology 89.1: 228. 52. G.V. Vidya Lakshmi, Y. Vasanthi, A. Suneetha, M. Nagaraju, (2020),” Imbalanced Data in Sensible Kernel Space With Support Vector Machines Multiclass Classifier Design”, Journal of Critical Reviews, Vol 7, Issue 4, pp: 820-824.

Journal of critical reviews 354