Scribd
Upload
90 views10 pages

Bofinal

This document discusses a handwritten text recognition software project that uses an artificial neural network approach. It begins with an introduction to optical character recognition and text recognition. It then discusses artificial neural networks and their structure. The main body of the document describes the image processing steps involved in text recognition, including pre-processing, feature extraction, and an example output screenshot. It concludes by discussing future applications and references used.

Uploaded by

Animesh Prasad
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
90 views10 pages

Bofinal

This document discusses a handwritten text recognition software project that uses an artificial neural network approach. It begins with an introduction to optical character recognition and text recognition. It then discusses artificial neural networks and their structure. The main body of the document describes the image processing steps involved in text recognition, including pre-processing, feature extraction, and an example output screenshot. It concludes by discussing future applications and references used.

Uploaded by

Animesh Prasad
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 10

1.

INTRODUCTION
This project “Handwritten Text Recognition” is a software algorithm project to recognize any
hand written text efficiently on computer with input is either an old optical image or currently
provided through touch input, mouse or pen.

Text recognition, usually abbreviated to optical character recognition or shortened OCR, is


the mechanical or electronic translation of images of handwritten, typewritten or printed text
(usually captured by a scanner) into machine-editable text. It is a field of research in pattern
recognition, artificial intelligence and machine vision. Though academic research in the field
continues, the focus on character recognition has shifted to implementation of proven
techniques. Optical character recognition is a scheme which enables a computer to learn,
understand, improvise and interpret the written or printed character in their own language,
but present correspondingly as specified by the user. Optical Character Recognition uses the
image processing technique to identify any character computer/typewriter printed or hand
written. A lot of work has been done in this field. But a continuous improvisation of OCR
techniques is being done based on the fact that algorithm must have higher accuracy of
recognition, higher persistency in number of times of correct prediction and increased
execution time.
2. ARTIFICIAL NEURAL NETWORK
An early phase of Neural Network was developed by Warren McCulloch and Walter Pitts in
1943 which was a computational model based on Mathematics and algorithm. This model
paved the way for research which was focused on the application of Neural Networks in
Artificial Intelligence.

Artificial neural network is basically a mesh of large number of interconnected cells. The
arrangement of cells are such that each cell receives an input and drives an output for
subsequent cells and each cell has a pre-defined.

The diagram below is a block diagram that depicts the structure and work flow of a created
Artificial Neural Network. The neurons are interconnected with each other in a serial manner.
The network consist of a number of hidden layers depending upon the resolution of
comparison of inputs with the dataset.
Fig. 2.1 Block Diagram of Artificial Neural Network

3. IMAGE PROCESSING INVOLVED IN TEXT


RECOGNITION

3.1 PRE-PROCESSING OF SAMPLE IMAGE

Pre-processing of the sample image involves few steps that are mentioned as follows :-

• Grey-scaling of RGB image :- Grey-scaling of an image is a process by which an RGB


image is converted into a black and white image. This process is important for
Binarization as after grey-scaling of the image, only shades of grey remains in the
image, binarization of such image is efficient.
• Binarization :- Binarization of an image converts it into an image which only have
pure black and pure white pixel values in it. Basically, during binarization of a grey-
scale image, pixels with intensity lower than half of the full intensity value gets a zero
value converting them into black ones. And the remaining pixels get a full intensity
value converting it into white pixels.
• Inversion :- Inversion is a process in which each pixel of the image gets a colour which
is the inverted colour of the previous one. This process is the most important one
because any text on a sample image can only be extracted efficiently if it contains only
one colour which is distinct from the background colour and it is only required if the
objects we have to identify if of darker intensity on a lighter background.
Fig. 3.1 RGB to Grey Scale to Binary to Inversion
3.2 FEATURE EXTRACTION
Features of a text depicts the morphological and spatial characteristics in the image. Feature
extraction is a method of extracting of features of texts from the sample image.

3.2.1 Statistical Feature Extraction

In this type of extraction the extracted feature vector is the combination of all the features
extracted from each character. The associated feature in feature vector of this type of
extraction is due to the relative positions of features in character image matrix.

3.2.2 Structural Feature Extraction

This is a primitive method of feature extraction which extracts morphological features of a


text from image matrix. It takes into account the edges, curvature, regions, etc. This method
extracts the features of the way text are written on image matrix.
4. SCREENSHOT OF IMPLEMENTATION
4.1 Output
5. FUTURE SCOPE
The application of this HCR algorithm is extensive. Now-a-days recent advancement in
technologies has pushed the limits further for man to get rid of older equipment which posed
inconvenience in using. In our case that equipment is a keyboard. There are many situations
when using a keyboard is cumbersome like :-

• We don’t get fluency with keyboard as real word writing.


• When any key on keyboard is damaged.
• Keyboard have scripts on its keys in only one language.
• We have to find each text on keyboard which takes time.
• In touch-enabled portable devices it is difficult to add a keyboard with much ease.
On the other hand if we use an HCR software in any device, we can get benefits like :-

• Multiple language support.


• No keyboard required.
• Real world writing style support.
• Convenient for touch enabled devices.
• Previously hand written record can be documented easily.
Extensive features can also be added to the software like :-

• Translation.
• Voice reading.
6. CONCLUSION
Classification of the texts and learning of image processing techniques is done in this project.
Also the scheme through which project is achieved is Artificial Neural Network scheme. This
work was basically focused on envisaging methods that can efficiently extract feature vectors
from each individual character. The method we came up with gave efficient and effective
result both for feature extraction as well as recognition. There are also different methods
through which ‘handwritten text recognition’ is achieved.
7. REFERENCES
[1] TensorFlow for Deep Learning : From Linear Regression to Reinforcement Learning 1st

Edition by Bharath Ramsundar.

[2] Fundamentals of Neural Networks 1st Edition by Laurene Fausett.

[3] www.geeksforgeeks.com.

[4] www.medium.com.

[5] www.datacamp.com

You might also like