We hereby declare that this project has been carried out by us under the supervision of Md. We also declare that neither this project nor any part of this project has been submitted elsewhere for the award of any degree or diploma. Syed Akhter Hossain, Head, Department of Computer Science and Engineering for his kind help in completing our project and also to other faculty members and staff of the Department of Computer Science and Engineering at Daffodil International University.
This application can detect and recognize Bengali text that is captured by a mobile device or selected from gallery and displays the recognized text on the phone screen. For developing this application, we have used Tesseract Optical Character Recognition Engine (OCR) Engine which is developed by Google is an open source OCR application. This project is useful for foreign tourists to navigate while traveling through this country.
This application allows users to perform many actions such as copying and changing text within minutes, instead of wasting time retyping it. The user interface of this application is user-friendly so that any user can use this application. An example of Optical Character Recognition (OCR) is ASCII code that the computer can manipulate.
The user interface will be very simple so that any user will be able to use this application very easily.
BACKGROUND
- Introduction
- Related Works
- Comparative Studies
- Scope of the Problem
- Challenges
But in our software we have used the most accurate OCR engine Tesseract which is developed by Google. Some authors addressed noise detection and cleaning phase in their works to make the system convenient. However, a comprehensive solution to eliminate all types of noise for Bengali OCR system is not available.
The reader has already understood that Bengali has not only basic characters; but also rich in modifiers and compound characters. Again, lack of standard or benchmark samples does not allow for extensive testing of their use. Existing applications are built for printed documents, where complex structure of documents is assumed not to occur.
Also, these applications do not have a quality page layout analyzer that can automatically recognize the image paragraph and text of the input image. These have some limitations in line segmentation for images of newspaper documents due to the occurrence of the problem of merging between two lines in digital images. Many existing applications work well but are not user-friendly and have a very complex user interface.
The characters in Bengali are not alphabetic like in English, where characters have mostly one-sound, one-symbol characteristics.
REQUIREMENT SPECIFICATION
- Business Process Modeling
- Requirement Collection and Analysis
- Use Case Modeling and Description
- Design Requirements
Machining follows a traditional step-by-step pipeline, but some of the stages were unusual in their time, and may still be today. The first step is a coherent component analysis where the contours of the components are saved. This was a computationally expensive design decision at the time, but had a significant advantage: by inspecting contour nesting and the number of child and grandchild contours, inverted text is easily detected and recognized as easily as black-on-white text.
Because the adaptive classifier may have learned something useful too late to make a contribution near the top of the page, a second pass across the page is performed in which words that were not recognized well enough are re-recognized. The row-finding algorithm is one of the few parts of Tesseract that has already been published. The mid-height is roughly the size of the text in the region, so it's safe to filter out blobs smaller than a fraction of the mid-height, as they are most likely punctuation marks, diacritics, and noise.
The final step of the line creation process joins blocks that overlap at least halfway horizontally, placing diacritical marks together with the correct base, and correctly associating parts of some broken characters. Tesseract solves most problems by measuring gaps in a limited vertical range between the baseline and midline. Part of the recognition process for any character recognition engine is to identify how a word should be segmented into characters.
Candidate cap vertices are found from concave vertices of a polygon approximation of the outline, and can either have another concave vertex opposite, or a line segment. Any chop that fails to improve the confidence of the result is undone but not completely discarded so that the chop can be reused later by the associate if necessary. The associator makes an A* (best first) search of the segmentation graph of possible combinations of the maximum chopped patches in candidate characters.
When A* segmentation search was first introduced around 1989, Tesseract's accuracy on broken characters was well ahead of commercial engines of the time. Since Google has already made these files open source and these files are also verified, we used these files to build our system. After pressing Start OCR, the recognized text, which is the result of our system, will be displayed.
DESIGN SPECIFICATION
Front-end Design
Frame Layout is designed to block an area on the screen to display a single item in the android application. In general, Frame Layout should be used to hold a single child view, because it can be difficult to arrange child views in a way that is scalable to different screen sizes without the children overlapping each other. We use Frame Layout to block the loaded image area on the screen to display the selected image to the user, which is important in our application to confirm that the image is selected perfectly.
Because a Fragment is a part of the activity that enables more modular design of the activity in the application.
Back-end Design
In figure 4.1 you can see that real users who use this application have given their opinion about this application. Here the opinion of 19 real users is available, most of which means that 17 users voted very satisfied after using this application.
Implementation Requirement
Users
IMPLEMENTATION AND TESTING
- Implementation of Database
- Implementation of Front-end Design
- Implementation of Interactions
- Testing Implementation
- Test Results and Reports
On the other hand, if the user clicks on the gallery, the phone gallery will open and he/she can select the picture from the storage. To evaluate the compliance of the system with the specified requirements is the purpose of this test. 03 To check all options are working or not 04 To check if media permission is requested or not 05 To check if camera option is working or not 06 To check if gallery option is working or not.
When we tested this application by real-time users, we asked these questions to each user. 03 To check that all options are working or not. 04 To check whether media permission has been requested or not. 05. To check whether camera setting is working or not. 06. To check whether gallery setting is working or not. 07. To check whether the OCR engine is working or not. not successful 08 To check if image change stopped it or not. 09 To check whether output text is editable or not. 10 To check whether output text can be copied or not. We found the test results quite successful. 07 To check whether the OCR engine is working or not 08 To check whether image modification stopped it or not 09 To check whether output text is editable or not.
CONCLUSION AND FUTURE SCOPE
Discussion and Conclusion
Limitations
Scope for Further Developments
1] More information about Tesseract (software), available at https://en.wikipedia.org/wiki/Tesseract_(software), last accessed November 25, 2017 at 11:30 PM. 2] Learn more about Android, available at https://www.tutorialspoint.com/android, last accessed December 1, 2017 at 10:20 PM. 3] Learn more about Android development, available at https://developer.android.com/develop/index.html, last accessed on February 25, 2018 at 12:45 PM.
4] Learn more about Android Studio, available at https://developer.android.com/studio/index.html , last accessed on March 22, 2018 at 09:17. 7] Pick image from gallery or camera on Android, available at http://droidmentor.com/pick-image-from-gallery-or- camera/ , last accessed on February 20, 2018 at 3:22 am.
Plagiarism
