You are on page 1of 20

Image processingMeaning retrieval of

word in an image.
Under the guidance of:
S. Naveen kumar
(Asst.Professor)

Contents:

Aim
Introduction
Software used
Block diagram
Pros and cons
Applications
Work division
Conclusion

Aim
The input for this application will be an image of a word .
The relative meaning of that word will be retrieved from
the database and displayed.

Introduction
The whole world is now wanting accuracy with speed.
'Image sensing and processing' is one of those techniques
to achieve this goal.
QR scanning, Image cryptography, security issues are some
of the applications. The input for this technique is an image
and the output will be a processed image or information
related to that input image.

Software required
MATLAB

The whole process is divided into three phases:


Image detection using Mat Lab.
Creating a Data Base of meanings.
Connecting the output of Processed Image with the
DataBase and retrieving the meaning of the word.

START

CAPTURE
AN IMAGE
OF A WORD

DISPLAY THE
DATA AS THE
FINAL INPUT

TERMINATE
THE
APPLICATION

ALLIGN THE
IMAGE OF
PROPERLY NOT
ALLIGNED

RETRIVE
CORRESPONDING
DATA FROM THE
DATA BASE

PROCESS
THE
IMAGE

MAKE THIS
TEXT AN
INPUT TO A
DATA BASE

IDENTIFY THE
REGION OF THE
IMAGE WHERE
TEXT IS PLACED

CONVERT THE IMAGE IN


TO EDITABLE TEXT

BLOCK DIAGRAM

Steps involved

Reading the image.


Segmentation.
Converting to binary image.
Trimming the image.
Normalization.
Classification.
Comparison with templates.
Text extraction.

Loading the image


Load the image into Mat Lab. If the image is not properly aligned, it
can be rotated in plane, but significant out of plane rotations may
require additional pre-processing.
Commands:
colorImage = imread('handicapSign.jpg');
figure; imshow(colorImage); title('Original image')

Next Step: Segmentation


The image is cropped to fit the text.
The noise in images is removed by cropping the part of the image
which has text After that, line by line is separated.
Example :

To convert any image into Binary image


The threshold for the binary image transformation is 0.99 (= bn
im2bw (image, 0.99)). This threshold was taken to colors with
close to 255 (maximum) RGB values are considered as 0 in the
binary image.

Trimming the image


The parts of the image which have text are to be trimmed. This is shown in the
next slide.
The next step is to separate each line.
Once obtained separately each line of the image, it is necessary to remove one
letter from the image matrix fl. For this function the bwlabel used, which label
connected components of the image. In other words, this function has the solid
lines and lists.
Example :

Normalization
Then each letter is normalized to a size of 42 x 24 pixels, which
is the size of the template that will perform the co-relation. This
technique is formally is known as normalization.

Templates
Each template is a binary image bmp 42 x 24 pixels. They are stored
separately so that they can be used for reference.

Final step
Then the image is compared with templates and most approximate
results are stored as outputs.
With this step, image processing part is done and the final text is
extracted is given as input the database for retrieval.

Real-time Applications
Image processing is a labor camp absolutely crucial for
those groups and industries that are working in areas such as
medical diagnosis, astronomy, geophysics, environmental
sciences, data analysis in laboratories, Industrial inspection, etc.
To sort out the posts according to their destination by reading
their address present on their envelope this makes sorting faster
and accurate

In banking services for reading the amount and name of the person
present on the cheque, DD etc
Old scripts can be stored digitally so that they can be reproduced when
ever needed.
The same principle can be used for number plate recognition and in the
name card reader.

Advantages
The main advantage is when ever there is huge
data in the form of image then the text in that
can be extracted easily

Limitations
Black colored text printed on white sheet is preferred for maximum
accuracy.
No extra light effects must be present while capturing the image.
Font style variation must be avoided through out the text.
It is difficult to differentiate between characters such as I&i, S&s, Z
&z.
Font of text taken as input must match with font of database images for
better accuracy.

Test Example

You might also like