A SURVEY ON OPTICAL CHARACTER RECOGNITION TECHNIQUES

 

A SURVEY ON OPTICAL CHARACTER RECOGNITION TECHNIQUES

ABSTRACT :

At present scenario, there is growing demand for the software system to recognize characters in a computer system when information is scanned through paper documents. This paper presents detailed review in the field of Optical Character Recognition. Various techniques are determined that have been proposed to realize the center of character recognition in an optical character recognition system. OCR (Optical Character Recognition) translates images of typewritten or handwritten characters into the electronically editable format and it preserves font properties. Different techniques for preprocessing and segmentation have been surveyed and discussed in this paper.

Keywords: Character Recognition System, Image Segmentation, OCR, Preprocessing, Skew correction, Classifier

1. INTRODUCTION

OCR (Optical Character Recognition) translates images of typewritten or handwritten characters into machine editable format. OCR reads damaged or low-quality codes and returns the best guess at what the code is. It is widely used as a form of information entry from printed paper data records, whether passport documents, invoices, bank statements, computerized receipts, business cards, mail, printouts of static data, or any suitable documentation. OCR does not deal with quality and sharpness of characters. To overcome the limitations of OCR a new approach comes into picture which is OCV. Projection Profile-based methods used makes segmentation easy to separate the text in document image into lines, words, and characters independent of the Language in the Text. Different methods are used at each intermediate stage of OCR. Text Segmentation is done using Projection Profile method. They proposed an algorithm for correction of the skew angle of the text document [1]. Blur is the important factor that damages OCR accuracy. In this paper prediction method based on a local blur estimation is proposed. The relation between blur effect and character size is investigated which is useful for the classifier. Classifier separates the given document into three classes: readable, intermediate, non-readable classes [2]. The grading system is used to evaluate the performance of printed text using various quality measures. The recognition results showed high recognition rate as the system was able to perform a recognition rate of 98.69 % along with a precision of 0.9857 and a sensitivity of 1 [3]. This paper presents complete OCR (Optical Character Recognition) system for camera captured image/graphics embedded textual documents for handheld devices [4]. Paper [5] describes the skew detection and correction of scanned document images written in Assamese language using the horizontal and vertical projection profile analysis OCR consists of many phases such as Pre-processing, Segmentation, Feature Extraction, Classifications and Recognition [6].

Existing system :

OCR (Optical Character Recognition) translates images of typewritten or handwritten characters into machine editable format. OCR reads damaged or low-quality codes and returns the best guess at what the code is. It is widely used as a form of information entry from printed paper data records, whether passport documents, invoices, bank statements, computerized receipts, business cards, mail, printouts of static data, or any suitable documentation. OCR does not deal with quality and sharpness of characters. To overcome the limitations of OCR a new approach comes into picture which is OCV. Projection Profile-based methods used makes segmentation easy to separate the text in document image into lines, words, and characters independent of the Language in the Text. Different methods are used at each intermediate stage of OCR. Text Segmentation is done using Projection Profile method. They proposed an algorithm for correction of the skew angle of the text document [1]. Blur is the important factor that damages OCR accuracy. In this paper prediction method based on a local blur estimation is proposed. The relation between blur effect and character size is investigated which is useful for the classifier. Classifier separates the given document into three classes: readable, intermediate, non-readable classes [2].

Proposed system :

The grading system is used to evaluate the performance of printed text using various quality measures. The recognition results showed high recognition rate as the system was able to perform a recognition rate of 98.69 % along with a precision of 0.9857 and a sensitivity of 1 [3]. This paper presents complete OCR (Optical Character Recognition) system for camera captured image/graphics embedded textual documents for handheld devices [4]. Paper [5] describes the skew detection and correction of scanned document images written in Assamese language using the horizontal and vertical projection profile analysis OCR consists of many phases such as Pre-processing, Segmentation, Feature Extraction, Classifications and Recognition [6].

Modules :

1.1 Digitization Digitization is the process of converting a paper-based handwritten document into electronic format. Here, each document consists of only one character. The electronic conversion is accomplished by using a method whereby a document is scanned and an electronic representation of the original document as an image file format is produced. The author used various scanners for digitization, and the digital image was going for next step that is a preprocessing phase.

1.2 Pre-processing In The pre-processing phase, there is a series of operations performed on the scanned input image. It enhances the image rendering it suitable for segmentation the gray-level character image is normalized into a window sized. After noise reduction, a bitmap image is produced. Then, the bitmap image was transformed into a thinned image.

1.3 Segmentation The Segmentation phase is the most important process. Segmentation is done by separation from the individual characters of an image. Segmentation of handwritten characters into different zones (upper, middle and lower zone) and characters is more difficult than that of printed documents that are in standard form. This is mainly because of variability in a paragraph, words of line and characters of a word, skew, slant, size and curved. Sometimes components of two adjacent characters may be touched or overlapped and this situation creates difficulties in the segmentation task. The touching or overlapping problem occurs frequently because of modified characters in upper-zone and lower-zone.

1.4 Feature Extraction and classification Feature extraction is the phase which is used to measure the relevant shape contained in the character. In the feature extraction phase, one can extract the features according to levels of text, e.g., character level, word level, line level and paragraph level. The classification phase is the decision making phase of an OCR engine, which uses the features extracted in the previous stage for making the class memberships in pattern recognition system. The preliminary aim of classification phase of OCR is to develop the constraint for reducing the misclassification relevant to feature extractions.

CONCLUSION

 “This paper elaborated survey of disparate techniques for OCR” has been studied. Handwritten character, natural scene images, business cards and TV set images are selected for experimentation. A systematic flow of OCR system is discussed. In this paper projection profile based method for segmentation, fourier transform technique is for pre-processing, and nearest neighbour classifier for classification are described. This paper can be helpful to the researcher for selecting most appropriate techniques to achieve optimum results for application according to a different parameter described in the previous section.

REFERENCES

 [1] A. S. Sawant, “Script Independent Text Pre-processing and Segmentation for OCR,” Int. Conf. Electr. Electron. Signals, Commun. Optim. - 2015, pp. 1–5, 2015.

[2] V. Kieu, F. Cloppet, and N. Vincent, “OCR Accuracy Prediction Method Based on Blur Estimation,” 2016 12th IAPR Work. Doc. Anal. Syst., pp. 317–322, 2016.

[3] J. B. Pedersen, K. Nasrollahi, and T. B. Moeslund, “Quality Inspection of Printed Texts,” IWSSP 2016- 23rd Int. Conf. Syst. Image Process. 23-25 May 2016, Bratislava, Slovakia, pp. 6–9, 2016.

[4] A. F. Mollah, N. Majumder, S. Basu, and M. Nasipuri, “Design of an Optical Character Recognition System for Camera- based Handheld Devices,” IJCSI, vol. 8, no. 4, pp. 283– 289, 2011.

[5] B. Jain and M. Borah, “A Comparison Paper on Skew Detection of Scanned Document Images Based on Horizontal and Vertical,” IJSRP, vol. 4, no. 6, pp. 4–7, 2014.

[6] E. N. Bhatia, “Optical Character Recognition Techniques : A Review,” IJARCSSE, vol. 4, no. 5, pp. 1219–1223, 2014.

[7] M. Shen, “Improving OCR Performance with Background Image Elimination,” 2015 12th Int. Conf. Fuzzy Syst. Knowl. Discov., pp. 1566–1570, 2015.

[8] A. Coates et al., “Text Detection and Character Recognition in Scene Images with Unsupervised Feature Learning.”

[9] P. Road, “Confidence Guided Progressive Search and Fast Match Techniques for High Performance ChineseEnglish OCR *,” IEEE, pp. 89–92, 2002.

[10] H. Wang and J. Kangas, “Character-Like Region Verification for Extracting Text in Scene Images,” no. 11, 2001.

[11] I. Kastelan, S. Kukolj, V. Pekovic, V. Marinkovic, and Z. Marceta, “Extraction of Text on TV Screen using Optical Character Recognition,” IEEE, pp. 153–156, 2012.

[12] J. Diaz-escobar, “Optical Character Recognition based on phase features,” IEEE, 2015.

[13] A. Thilagavathy, K. Aarthi, and A. Chilambuchelvan, “A Hybrid Approach to Extract Scene Text from Videos,” ICCEET, pp. 1017–1022, 2012.

[14] S. Goyal, “Optical Character Recognition,” IJARCSSE, vol. 3, no. 11, pp. 982–985, 2013.

[15] G. Vamvakas, B. Gatos, N. Stamatopoulos, and S. J. Perantonis, “A Complete Optical Character Recognition Methodology for Historical Documents,” pp. 525–532, 2008.

[16] L. S. Yaeger, B. J. Webb, and R. F. Lyon, “Search for Online , Printed Handwriting N EWTON,” Am. Assoc. Artif. Intell., vol. 19, no. 1, pp. 73–90, 1998.

[17] J. Hu, S. G. Lim, and M. K. Brown, “Writer independent on-line handwriting recognition using an HMM approach,” J. PATTERN Recognit. Soc., vol. 33, pp. 133–147, 2000.

[18] A. Funada, D. Muramatsu, and T. Matsumoto, “The Reduction of Memory and the Improvement of Recognition Rate for HMM On-line Handwriting Recognition,” IEEE, pp. 0–5, 2004.

[19] J. r´ı Matas, “Real-Time Scene Text Localization and Recognition,” IEEE, pp. 3538–3545, 2012.

[20] H. Lin and C. Hsu, “Optical Character Recognition with Fast Training Neural Network,” IEEE, pp. 1458–1461, 2016.

[20] C. N. E. Anagnostopoulos, I. E. Anagnostopoulos, V. Loumos, and E. Kayafas, “A License Plate-Recognition Algorithm for Intelligent Transportation System Applications,” IEEE, vol. 7, no. 3, pp. 377–392, 2006.

[21] Y. J. Zhang, “A survey on evaluation methods for image segmentation,” pp. 1–13.

[22] A. Singh, K. Bacchuwar, and A. Bhasin, “A Survey of OCR Applications,” Int. J. Mach. Learn. Comput., vol. 2, no. 3, pp. 314–318, 2012.

Author : ss infotech

Share this

Related Posts