Website Text Translation and Image Translation from a URL using Optical Character Recognition (OCR) I. Introduction e live in a world of information system at present. We are very dependent to various websites for information about almost everything. For this purpose, people all over the world goes through numerous websites every day. But all websites are not available in their native languages. Around 75% of the world's population does not speak in English according to BBC -UK report1 [2] . Here comes the need for translating the contents of the websites. Also, sometimes the images of the websites contain texts which are also need to be translated. Google translator is widely used for this translation purpose. It can translate texts of any websites using the url of the website. But it doesn't translate the texts inside the images of that website. If anyone searches for an educational website and there is an image of a notice, he/she will not be able to read it as it won't be translated using google translator. For image translation there are also many apps and websites which are widely used to translate the texts inside of an image to any desired language. But they only deal with images. OCR (Optical Character Recognition) is widely used for the image translation method. It is a technology that recognizes text within a digital image [4] . It is commonly used to recognize text in scanned documents and images [4] . In our work, we tried to create a platform where the users will be able to translate the whole website in any language using the URL and they will also be able to translate the image texts too. We have also created a platform which will convert random images where the numbers will also be translated from English to any languages. We worked with only Bangla and Arabic numbers here. But English numbers can also be translated to other language numbers too only by editing the algorithm we created. We organized this paper in this way: Section 1. Gives the introduction of our work, section. 2. Eplains the implementation details of our website, section. 3. Includes the three algorithms we created, section. 4. Presents the outcomes of the experiments, as well as a comparison to the current procedures and the last section. 5. Contains the conclusion and future work. Secondly, for image URL translation, when we put a website URL and run that in our website it also collects all the image URL that website has and show it at the end of our website. By selecting an image URL, we collect the image from google and convert it to word by using tesseract OCR and save it in a file. After that we call the .txt file and show it to our website. When we select an image URL, we see that .txt file along with the image and translation tool. Now we can translate the image and read the image text in any language. The figure 2 portrays the translation process of the text from an image url. We also show the image text in a format so that it is easy to understand and easy to read. # II. Implementation Details Lastly, for image translation, we take an image as an input to translate the image text along with numbers. We use our own algorithm to translate the image text and numbers as a sample we use English to Bengali or English to Arabic/Persian Language translation. When we put an image, it converts the image to text and put it in a .txt file. Then using a function to convert the numbers from English to Bengali or Arabic/Persian numbers and show the whole file in our website after converting the numbers and we get the following results in figure 3. And figure 4 shows the translated view of an image sample from English to Bengali and from English to Arabic respectively. # IV. Results Analysis This section describes the results of our works. In all three sectors of our work, we used the term Accuracy to calculate the performance of them. Accuracy: It is defined as the ratio of translated words and total words. Here w is the number of properly translated words, and W is the number of total words. # Accuracy = w / W We have experimented Up to 70 websites and up to 50 random images with our approach. Our website reaches almost 83 percent accuracy in the field of website translation and 85 percent in the field of translation of images from those websites. The accuracy of the translation of the random images from English to Bangla reaches 98 percent and from English to Persian it reaches almost 93 percent. We tested this approach with various text fonts, and our website accurately translated them all. # V. Conclusion and Future Work We have implemented an easier and userfriendly website which takes an url as input and translate the website in any desired language. Using this platform, users will be able to get the information of any website in their comfortable language and it is also timesaving as it translates any website using only a URL. It also translates the texts inside the images of the website. The users are also able to extract the texts of a random image and we have used our own algorithm to translate the English numbers to Bangla and Arabic numbers. In future, this paper will be helpful to build a mobile application where one can add camera module to take an image and translate it through the app where they can translate numbers too. This paper can also help to build an app or a website that will be able to take any url from any barcode and translate both the text and images. In future this paper will be helpful to create a new algorithm to translate text of all the images of the website in a single webpage along with the web text just like the original website. 1![Figure 1: The homepage of the website and the website url to be tested](image-2.png "Figure 1 :") 34![Figure 3: View of the image insert module of the website](image-3.png "Figure 3 :Figure 4 :") 5![Figure 5: Algorithm to Image Translation (English to Bengali)](image-4.png "Figure 5 :") 6![Figure 6 describes the algorithm to translate the text of a random image from English to Persian.](image-5.png "Figure 6") 6![Figure 6: Algorithm to Image Translation (English to Persian) Figure 7 describes the algorithm to translate the text of an image from a website url from English to any language.](image-6.png "Figure 6 :") 7![Figure 7: Algorithm for URL Translation](image-7.png "Figure 7 :") © 2022 Global Journals * An Image Processing Approach to Linguistic Translation ShruthiKubatur SuhasSreehari RajeshwariHegde December 2011 Bangalore, India Dept. of Electrical & Computer Engg, University of Windsor, Windsor, Canada Dept. of Telecommunication Engg, B.M.S. College of Engineering * Tourist's Translator based on Digital Image Processing and Hybrid Translation RijwanKhan AryanKaushal AyushAgarwal AvdheshKumar International Journal of Innovative Technology and Exploring Engineering (IJITEE) 2278-3075 9 March 2020 * Extraction of Text from an Image and its Language Translation Using OCR GRHemalakshmi MSakthimanimala JSalai Ani Muthu International Journal of Engineering Research in Computer Science and Engineering (IJERCSE) 2394- 2320 April 2017 * Text detection in natural and computer-generated images MandanaAzmi Can Özgen Fasounaki 2017 Hazim Kemal Ekenel * An overview of the Tesseract OCR engine RSmith Ninth International Conference on 2007. 2007. 2007. September 2 Document Analysis and Recognition * English to Spanish translation of signboard images from a mobile phone camera ACanedo-Rodriguez SKim JHKim YBlanco-Fernandez 10.1109/SECON.2009.5174105 IEEE Southeastcon 2009. 2009 * Optical Character Recognition for printed Tamil text using Unicode RSeethalakshmi TRSreeranjani TBalachandar AbnikantSingh MarkandeySingh RitwajRatan SarveshKumar Journal of Zhejiang University SCIENCE 1009-3095 2005 * FLD based Unconstrained Handwritten Kannada Character Recognition SK VijayaKumar International Journal of Database Theory and Application December 2000 * A technical review on text recognition from images Pratik Madhukar Manwatkar Dr RKavita Singh IEEE Sponsored 9th International Conference on Intelligent Systems and Control (ISCO) 2015