Verification of documents on the web resource using the method of text recognition

DOI: 10.31673/2412-9070.2022.035159

Abstract

The article is devoted to the coverage of current issues related to the verification of documents on the web resource using text recognition methods. It has been established that there are various methods and software solutions for digital identity verification. It is important to introduce artificial intelligence technologies and the possibility of refinement for individual tasks, which іі increases business efficiency many times. An important drawback is the deployment of platforms on servers, due to which the financial cost of integration and renting the necessary software increases. This article discusses the verification method of checking identity documents, which plays an important role in the processes of opening new bank accounts, signing and concluding financial agreements, using artificial intelligence technologies and authentication methods A method has been developed that uses the method of integration in the form of an API. The «Tesseract» module for JavaScript was used, a module was developed for a better and faster verification method, divided into several blocks of finding coefficients to confirm the liquidity of the document. An algorithm for finding coefficients to confirm the liquidity of a document and a mathematical model of document authentication have been developed. When using the built-in camera on a mobile or portable device, as well as artificial intelligence technology and the proposed authentication algorithm, it is possible to analyze the image in order to obtain an authenticity assessment to determine whether the identification document is fake or genuine. It has been established that this method prevails due to lower financial costs in maintenance and installation, the technology does not require additional servers or software installation. Also, the proposed technology speeds up the verification process and its accuracy, working only with documents.

Keywords: verification; authentication; artificial intelligence; API; JavaScript; Rusted Identity Network.

References
1. Wang T., Wu D. J., Coates A. End-to-End Text Recognition with Convolution Neural Networks // IEEE Conf. Pattern Recognition. 2012. Р. 3304–3308.
2. Zhu Y., Sun J., Naoi S. Recognizing Naural Scene Characters by Convolutional Neural Network and Bimodal Image Enhancement // Workshop on Camera-Based Document Analysis and Recognition. 2012. Р. 69–82.
3. Chen X., Yang J., Zhang J. Automatic Detection and Recognition of Signs from Natural Scenes // IEEE Image Processing. 2004. vol. 13, no. 1. Р. 87–99.
4. Васильєв В. Н., Гуров І. П., Потапов А. С. Математичні методи і алгоритмічне забезпечення аналізу та розпізнавання зображень в інформаційно-телекомунікаційних системах. 2008. С. 22–28.
5. Koo H., Kim D. H. Scene Text Detection via Connected Component Clustering and Non-text Filtering // IEEE Processing. 2013. vol. 22, no. 6. Р. 2295–2304.
6. Hanif S. M., Prevost L. Text Detection and Localization in Complex Scenes using Constrained Adaboost Algorithm // IEEE Conf. Document Analysis and Recognition. 2009. Р. 1–5.
7. Mosleh A., Bouguila N., Hamza A. Image Text Detection Using a Bandlet-Based Edge Detector and Stroke Width Transform // British Machine Vision Conference. 2012. Р. 1–2.
8. Rosset, Zhu, Hastie. Boosting as a Regularized Path to a Maximum Margin Classifier // Journal of Machine Learning Research. 2004. №5. Р. 941–973.
9. Garcia C., Apostolidis X. Text Detection and Segmentation in Complex Color Images // IEEE Conf. Acoustics, Speech and Signal Processing. 2000. Р. 2326–2330.
10. Karatzas D., Antonacopoulos A. Text Extraction from Web Images Based on a Split-and-Merge Segmentation Method Using Colour Perception // IEEE Conf. Pattern Recognition. 2004. Р. 634–637.
11. Lienhart R., Kuranov A., Pisarevsky V. Empirical Analysis of Detection Cascades of Boosted Classifiers for Rapid Object Detection // MRL Technical Report. 2002. Р. 12–15, 17.
12. Hunspell spell chacker dictionary [Електронний ресурс]. URL: http://hunspell.sourceforge.net/.
13. Jaderberg M., Simonyan K. Synthetic data and artificial neural networks for natural scene text recognition. NIPS Deep Learning Workshop, 2014.
14. Lucas S. M., Panaretos A. ICDAR 2003 robust reading competitions: entries, results, and future directions. IJDAR, 2005. Р. 105–122.
15. Krizhevsky A., Sutskever I. Imagenet classification with deep convolutional neural networks. NIPS, 2012.