International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056

Volume: 05 Issue: 04 | Apr-2018 www.irjet.net p-ISSN: 2395-0072

Survey Paper: Image Reader For Blind Person

Saurabh Dokhe1, Mayuri Dube2, Sonal Gade3, Prof. Vidya Nemade4

1, 2, 3 ,4Department of Computer Engineering, G.V. Acharya Institute of Engineering and Technology

Mumbai University, Mumbai, 400098, Maharashtra, India.
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Visual impairment and illiterate, or have a like a topographic map, the brightness of each point denote
learning disability is one of the biggest drawback for the its height, and finds the lines that run along the at most
humanity, especially in this day and age when information and or top of ridges. Optical character recognition (OCR) is
people is interconnected a lot by text messages (electronic and convert our image i.e. captured images of printed text into
paper based) rather than talking [4]. There is a need for a unicode. This project is also useful for visually impaired
convenient text reader that is reasonable and readily available people who cannot read Text document illiterate, or have a
to the blind community [2]. This work In this research these learning disability. Optical Character recognition is used to
images are converted into audio output. It is mainly used in identify texts that have been created with non-computerized
the field of research in Character recognition, Artificial system. OCR makes it possible to apply techniques such as
intelligence and computer vision. In this research, as the machine translation, text-to- speech and text mining to the
recognition process is done using OCR and opencv library.it capture page. The final standard text document is served to
recognizes character using watershed algorithm and python the output devices. The output device can be a headset
programming. this paper describes the design, implementation connected to the speaker which can speak out the text
and experimental results of the device. This device consists of document a loud. Extra hardware is not required for
three modules, image processing for object count, cropping of accepting output. In our implementation we also set the
image and reading an image. volume speed, sound and rate of sound, so external headset
and speaker is not necessary. The anticipated system has
Key Words: OpenCV, TTS, OCR based book reader, been totally developed for working in CLI environments. The
Watershed, algorithm, python programming, computer vision anticipated system has been totally developed in Python
using PyCharm/PyDev IDE and PDB as an Debugger.
Meanwhile Computer Vision libraries were used for
1. INTRODUCTION providing out of the box capabilities to the system. NumPy,
SciPy, Speech Recogniser, OCR, PyTesseract, and other
Near-sighted participants and illiterate people report several technologies and libraries are used.
difficulties for understanding printed text or captured text
using current technology, including problems with Features:
alignment, focus, precision, mobility and efficiency [2]. We
present software that supports the visually impaired which 1.Interactive Command & Control Panel – The person
effectively and efficiently reads printed and captured text of will able to interact with system using the interactive voice
specific font. The proposed project uses the watershed based control panel providing commands for executing
algorithm that can be used by people to read Text document. various operations available through the system.
The project is on implementing image font .The design is
inspired by earliest studies with visually impaired people, 2. Image to Speech - Using the app one can able to read an
and it is moderate and portable, which enables a more image using OCR technology and convert it to speech using
efficiency. In this project we have developed a code to read opencv libraries available in core STL.
text using command prompt for challenged and illiterate.
The proposed fully included system has a camera which 3.Image Counting – Using image processing and
captures the image or you feed the printed text or image morphological operation along with computer vision
reading or object counting process. The faded image libraries we will able to detect the total number of objects in
internally work with software module i.e. OCR and opencv an image and provide so that based on heuristic user will
libraries. The techniques involved in segmenting the image, able to recognize the components in an image.
and merging contiguous regions based on their altitudinal
distribution of colour/intensity. In order to recognize 4. Automatic Bounding Region Cropping and Re-Scaling of
important regions, or text part of image. If the specified size Text Regions using computer vision.
contains all the important regions, we simply crop the source
image, to do the character recognition, image cropping and
object counting. Most of the access technology tools built for
people with blindness and limited vision and illiterate . a
watershed algorithm convert our image into gray scale . The
watershed transformation treats the image it operates upon

© 2018, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 1738
 International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 05 Issue: 04 | Apr-2018 www.irjet.net p-ISSN: 2395-0072

2. LITERATURE SURVEY

Table no:1

Sr. No. Title Technology used Result

1 Text to Speech Conversion Raspberry Pi, OCR A performance that is high enough and a
Using OCR readability tolerance of less than 2%, with the
Technique in Raspberry average time processing.
Pi[1] This portable device does not require internet
connection.
This portable device does
not require internet
connection.
2 Image to speech conversion Raspberry Pi, OCR, TTS It gives good quality input to the OCR using pi
for visually impaired[2] camera

3 Vision based Text Raspberry Pi, Due to the less resolution of the webcam, the
Recognition using OTSU thresholding, output obtained is not 100% accurate. The
Raspberry Pi[3] OCR system fails to extract
the foreground when they
possess a complex
background.
4 camera based label reader OCR, label reader This method can effectively separate the objects
for blind people [4] of interest from complex background. OCR is
used to perform
word recognition.
5 Text to Speech for the TTS & OCR It captures the text and gives good quality of
Visually MODI algorithm audio.
Impaired[5]
6 Camera based label reader Raspberry Pi& TTS & We have used the ADA boost algorithm in
for blind people[6] OCR, ADA Boost Raspberry PI Model for the conversion so that
algorithm the blind can easily understand and read the data

This paper presented their project for text detection and hospital signs, and bus numbers –as well as more variable
converts into audio format. Testing of device was done on forms such as shop signs, house numbers, and billboards.
raspberry pi platform. The R-pi is initially connected to the
internet through VLAN. The software is installed using Here we are using AdaBoost Algorithm for treating the visual
command lines. The first setup is to download the information and converting into audio speech [4]. The
installation script, second command is to convert it to proposed system helps visually impaired, illiterate, or have a
executable form and the last command starts the script learning disability to read product the project aims to
which does the rest of the installation work [1].The paper implement a reading aid that is small, lightweight, efficient,
says they proposed a device to help people with visual cost effective and of course user-pleasant the Raspberry Pi-
impairment. In this project, we developed a device that based system can be equipped with a high- resolution
converts an image’s text to audio format. This webcam the microcontroller-built system is easier to use
implementation required hardware. The basic framework is when compared to the mobile one. However, the accuracy of
this implemented system that captures an image, extracts the mobile in the conversion efforts is better, mainly due to
only the region of interest (i.e. region of the image that the high-resolution camera built in the device. Developing
contains text) and converts that text to audio. It is developed technology and in future expansions of this project, the R-Pi
using a Raspberry Pi and a Raspberry Pi camera [2]. This based system can be provided with a good and high-
project presents a sample system for recognition of text resolution camera contrasted with the one used in this
present in the image using raspberry pi. The system agenda project, and we anticipate, this will improve its inevitability.
consists of five well-designed components: Image We predict more work will be produced in this critical area
acquisition, Image pre- processing, Text extraction, Text to of assistive technology, and project that future transportable
speech conversion and Speech output [3]. This paper gadgets will have easy to use and built in mechanism as
proposed a system using Raspberry Pi Model for perusing reading assistances for the blind, similar, to the mobile-
the images from stereotypical forms – such as street signs, based solution presented here. Users should capture image
and then system read out the text from image. It will be more

© 2018, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 1739
 International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 05 Issue: 04 | Apr-2018 www.irjet.net p-ISSN: 2395-0072

applicable for persons those are going through visual [6] R.Mohanapriya, U.Nirmala,C.Pearlin Priscilla Smart
surgery. It can be suitable for road side text recognition so vision system for blind international journal of engineering
that visually impaired person can travel alone [5]. this paper and computer science issn:2319-7242 volume 3 issue 5, May
proposed a system using Raspberry Pi Model for scanning 2014
the images from stereotypical forms – such as street signs,
hospital signs, and bus numbers –as well as more variable [7] Sanjana.B, J.RejinaParvin Voice assisted text reading
forms such as shop signs, house numbers, and billboards. system for visually impaired persons using TTS method iosr
Here they are using AdaBoost Algorithm for processing the journal of vlsi and signal processing (iosr-jvsp) volume 6,
visual information and converting into audio with default issue 3, May 2016
sound.[6]

3. CONCLUSIONS

We have successfully implemented conversion of text image

into audio format. We also developed technique for object
detection in an image and cropping of textual part of image
using OpenCV libraries. Our algorithm successfully processes
the image and reads it out clearly with set volume. And also
detect the object and crop the text part of image. This is an
efficient as well as helpful device for the visually impaired,
illiterate, or have a learning disability people. We have
applied our algorithm on many images and found that it
successfully does its conversion.

4. FUTURE SCOPE

It can be used in blind schools and colleges. This can also be

used as application of artificial intelligence.it is helpful for
illiterate people. We also can make mobile application of
same topic

REFERENCES

[ ] K Nirmala Kumari, Meghana Reddy Image Text to

Speech Conversion Using OCR Technique in Raspberry Pi
International Journal of Advanced Research in Electrical,
Electronics and Instrumentation Engineering (An ISO 3297:
2007 Certified Organization) Vol. 5, Issue 5, May 2016

[2] Asha G. Hagargund, Sharsha Vanria Thota, Mitadru Bera,

Eram Fatima Shaik Image to speech conversion for visually
impaired Volume 3 - Issue 06 June 2017

[3] Nagaraja L, Nagarjun R S, Nishanth M Anand Vision

based Text Recognition using Raspberry Pi ) National
Conference on Power Systems & Industrial Automation
(NCPSIA) 2015

[4] A. SUBBIAH*, T. ARIVUKKARASU, M. S. SARAVANAN, V.

balaji camera based label reader for blind people Int. J.
Chem. Sci.: 14(S3), 2016, 840-844 ISSN 0972-768X

[5] Mrs.Shilpa Reddy K, Mounika S.K,Pooja K , Sahana N Text

to Speech for the Visually Impaired International Research
Journal of Computer Science (IRJCS) ISSN: 2393- 9842Issue
05, Volume 4 , May 2017

© 2018, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 1740