1. DEPARTMENT OF
COMPUTER SCIENCE AND ENGINEERING
MINI PROJECT REVIEW
BATCH NO:A12
20891A0539- MD WASEEM AKRAM
20891A0530- KHAJA MOINUDDIN
20891A0532- M SAI SACHITANANDA REDDY
0
OPTICAL CHARACTER RECOGNITION
UNDER THE GUIDANCE OF
Mr.S KRANTHI REDDY
Associate Professor
(CSE)
2. INTRODUCTION
Optical Character Recognition (OCR) is the process thatconverts an image of text into a machine-
readabletext format. For example,if you scan a form or a receipt, your computer saves the scan
as an image file.
You cannot use a texteditor to edit, search, or count the words in the image file. However,you can
use OCR to convert the image into a textdocument with its contents stored as text data.The
service aims to analyzea physical document's contentand convert the elementsinto a script that
can subsequently be utilized for processing data.For example,consider postaland mail sorting
services.
Artificial Intelligenceanalyzes the image's dark portions to recognize characters and numerals.
Typically,AI uses one of the followingapproaches to target one letter,phrase, or paragraphat a
time. PatternRecognition Technologiesuse a range of language,text formats, and handwritingto
train the AI system.
The program compares the letters on the detected letterpicture to the notes it has already
learnedto find matches.Feature Recognition algorithmuses rules based on specific character
properties to recognize new characters. Theamount of angled,crossing, or curved lines in a letter
is one exampleof a feature.
3. ABSTRACT
The"OpticalCharacterRecognition"isthe processof detectingtext
content on imagesandconvertingit to machine-encodedtext thatwe can
accessandmanipulate.
The "OpticalcharacterRecognition"project canautomaticallydigitize
printedorhandwrittendocumentsandextracttext content from them.
Thesystem iscapable of recognizingvariousfonts, languages,and
handwritingstyles. The extractedtext canthen be savedin a digital
format,makingit searchableandeditable.
4. LITERATURE SURVEY
The Optical CharacterRecognitionfocuses on extractingTextfrom Imageswith
specific Implemenatationof AdaptiveThreshold apparoach.The methodsin this
projectincludes screening of datasetsfromInternet. The majorfeatureof this
system isto extractanddisplay text withits informationin text formatfrom the
imageformat.Varioussources of imagewithtext areavailablelike: imagewith
captiontext, imagewithshadedbackground,some imageswithtext arealso
differentwith their color, alignment sizeetc. This variationmakesthe problem
very difficultto drawautomatically.Imageacquisitionis aprocesswhere an
imageisgiventhat isalreadyindigitalform.This isthe stagewherethe image
underconsiderationis taken.Imageacquisitioncontainsvariationin the
intensitylevels along the image.So many noisesarealso addedto the image.So
the firsttaskis requiredto denoisethe image.It’scalled pre-processing.The
proposed system used documentedimageasinput.
5. Thresholding is requiredfor imagesegmentation.It distinguishesthe imageregionsasobjects
or thebackground.Although the detected edgesareconsistof text edgesandnontext edgesin
everydetailcomponent sub-band,it distinguishesthem due to the factthatthe intensity of the
text edgesis higher thanthat of the non-text edges.An appropriatethreshold value is selected
andpreliminarilyremoved the non-text edgesin the detailcomponent subbands. Adaptive
thresholding employs to calculate the targetthreshold value. Thresholding processis applied
to discardunwanted pixel regionsthatareless thana calculatedthreshold value. The proposed
system used documentedimagesorscannedtext book pagesasinputs andproducedbetter
results. Currenttechnologies do not workwell for digitalimagesthatcontainingbent or arched
text or imagethat containspicturesmore thantext. It also unable to remove undesirabletext
from images.In contrast,the result is moreaccuratewithsmall number of unwanted
charactersfor normaldocumentedimageaswell aswith the scannedpages.This system also
introduceda new platform for text informationextractionin Linux operatingsystem. The future
plan is to implement anintelligent informationextractionsystem thatwill automaticallyskip
all unwantedcharactersor symbols.
6. EXISTING SYSTEM:-
There areseveral existingsystems andsoftwarefor Optical CharacterRecognition(OCR), which is
the technology used to convert printed orhandwrittentext into machine-readabletext.
Herearesome well-known OCR systems:
TesseractOCR: Tesseractisone of the most popular open-source OCR enginesdeveloped by Google.
It supportsmultiple languagesandcan be integratedintovarious applications.
ABBYYFineReader:ABBYYFineReaderis acommercialOCR softwarethat provideshigh accuracy
andsupportsmultiple languages.It'swidely used forconvertingscanneddocuments, PDFs, and
imagesinto editableformats.
AdobeAcrobathas built-in OCR functionality that allows you to convert scanneddocuments and
imagesinto searchableandeditable PDFs.
Microsoft Office OCR: MicrosoftOffice applicationslike WordandOneNote offerOCR capabilitiesto
convert imagesandPDFs into editabletext.
7. PROPOSED SYSTEM
Designinga proposed OCR system involvesconsideringvarious components and steps to achieveaccurate and efficient
text extraction from images.
Here's a general outline of a proposed OCR system:
Image Preprocessing:
Noise Reduction:Apply filtersto remove noise, artifacts,and background interferencefrom the input image.
Binarization:Convert the image to binary format (black and white) to enhancetext visibility.
Deskewing:Correct any skew in the image caused by scanningor image capture.
Use a suitableOCR library or engine likeTesseract,ABBYY, or others.
QualityAssurance:
Confidence Scores: Assignconfidencescores to each recognized characteror word to gauge accuracy.
Error Correction: Implement error correction mechanisms basedon the confidencescores to fix misrecognized
characters.
Remember that the effectivenessof an OCR system depends on the qualityof inputimages, the chosen OCR engine,
preprocessingtechniques,and the complexity of the content beingrecognized.
8. SYSTEM REQUIREMENTS
HARDWARE REQUIREMENTS
Processor: Intel i3 or more
RAM: 4 GB or more
HARD DISK: 256 GB
SOFTWARE REQUIREMENTS
Python Programming Language
Pycharm community software
Various InstalledModules such as OpenCV, pytesseract
9. SYSTEM ARCHITECTURE
The system architectureof an OpticalCharacter Recognition (OCR) project typically involvesa
seriesof interconnectedcomponents and processes designedto convert scannedor digital
images containingtext into machine-readabletext. Below is a high-leveloverview of the typical
system architecturefor an OCR project:
1. InputDataAcquisition:
•Scanneror ImageSource: Thiscomponent captures the physicaldocument or image containing
text. It could be a scannerfor physicaldocuments or an image filein various formats (e.g., JPEG,
PNG) for digitalimages.
2. Preprocessing:
•ImagePreprocessing: Before OCR can be applied,the inputimage may undergo several
preprocessingsteps to enhance its quality and readability.Common preprocessingtechniques
include noise reduction, image binarization(conversion to black and white), deskewing
(alignment correction), and removal of artifacts.
3. TextDetectionandLocalization(Optional):
•In some OCR systems, an additionalstep is includedto locate and identifyregions of interest
(ROI) within the image that contain text. Thisis particularlyusefulwhen dealingwith complex
documents containingmultipletypes of content.
10. SYSTEM ARCHITECTURE
4. OpticalCharacterRecognition(OCREngine):
FeatureExtraction: The OCR engine extracts features from the preprocessed image, such
as character shapes, patterns, and spatial relationships.
CharacterRecognition: The core of the OCR system, where the extracted features are used
to recognize individual characters and symbols.
Wordand TextLine Recognition: In addition to character recognition, OCR systems often
recognize complete words and text lines to improve accuracy and context understanding.
5.Text Output:
The OCR system generates machine-readable text from the input image. This can be output
in various formats, such as plain text, rich text (RTF), or PDF, depending on the project
requirements.
6.Security(Optional):
Implement security measures to protect sensitive data, especially if the OCR project
involves handling confidential or personal information.
16. TheOptical CharacterRecognition(OCR)projecthas successfully addressed
the need for automatedtext recognitionandextractionfrom scanned
documents, images,andother textual sources. This projectaimedto improve
dataaccessibility,efficiency,and accuracyin variousdomains,including
document management,dataextraction,andarchivalprocesses. The OCR
system implemented in thisprojecthas proven to be an effectivesolution for
processinga widerangeof documents. It can efficiently convert scanned
documents intomachine-readabletext, makingit easierto store, search,and
analyzetextual content. The OCR projecthas been successful in achievingits
primaryobjectivesof automatingtext recognition,enhancing data
accessibility,andimprovingdocument managementprocesses. The system's
configurabilityandscalability makeit a valuable assetfor our organization,
andongoingeffortswill ensure thatit remainsa reliable solution for the
foreseeable future.
CONCLUSION
17. The field of OpticalCharacterRecognition (OCR) is continuallyevolving, driven by advancements
in technology and increasingdemands for accuracy and versatility.Here are some potential
future enhancementsand developments in OCR technology:
Integrationof machine learningand deep learningtechniquesto enhance OCR accuracy,
especiallyfor complex fonts, languages,and challengingdocument layouts.
Better recognition of document layouts, tables, and hierarchicalstructures to improve text
extraction accuracy and preserve document formatting.
Development of real-timeOCR solutions for instanttext recognition from livevideo streams,
handwrittentext, or augmented realityapplications.
Enhanced mobile OCR applicationsfor smartphones and tablets,enablingusers to extract text
from images, businesscards, and documents on the go.
These future enhancementsin OCR technology aim to make text recognition more accurate,
versatile,and accessibleacross various industriesand applications.As technology continues to
advance, OCR will play a vitalrole in improving data extraction, document management, and
information accessibility.
FUTURE ENHANCEMENTS