1. extracted features. After that, a target logo

 

1.  
Introduction to the Proposed Research Work with Problem Identification

Logo sometimes also known as trademark have high
importance in today’s marketing world, because it carries the goodwill of the
company and the product. Logo recognition has been specifically used in
application areas such as enterprise identification, entertainment advertising,
vehicle recognition, road sign reading, and website summarization by image
content based image retrieval. For a particular logo recognition system,
features related to visual contents are first extracted to describe the logo
images. Then a similarity measure is defined to compare the query image with the
target images which are saved in a logo database using the extracted features.
After that, a target logo most similar to the query image is retrieved. Since
the query logo image may be taken by a hand held mobile phone-camera operating
at varying different viewpoints under different lighting environments. The
query image may differ substantially from the database target image due to
geometric transformations like viewpoint, rotation, and scaling changes and
photometric transformation like lighting, noise, and image blur.

We Will Write a Custom Essay Specifically
For You For Only $13.90/page!


order now

 

In current trends the logos
are playing a vital role in industrial and all commercial applications.
Fundamentally the logo is defined as it’s a graphic entity which contains
colors textures, shapes and text etc., which is organized in some special
visible format. But unfortunately it is very difficult thing to save their
brand logos from duplicates. In practical world there are several systems
available for logo reorganization and detection with different kinds of
requirements. In some partial occlusions it should be robust to transfer
the large range of photometric and geometric features of a logo which they are
not captured in isolation. Two dimensional global
descriptors are used for logo matching and reorganization. The concept of Shape
descriptors based on Shape context and the global descriptors are based on the
logo contours. There is an algorithm which is implemented for logo detection is
based on partial spatial context and spatial spectral saliency (SSS). The SSS
is able to keep away from the confusion effect of background and also speed up
the process of logo detection. All such methods are useful only when the logo
is visible completely without noise and not subjected to change. These types of
methods are not suitable for practical images where insufficient resolution is
the drawback of these methods. To overcome these drawbacks we proposed a
multiple descriptors method along with context dependent similarity concept.

 

 2.  Problem Definition

 

Therefore it is a challenge to the logo
recognition system to extract the features robust to the above inevitable
imaging variations. A logo is nothing but a graphic entity containing colors,
shapes, textures, and perhaps text as well, organized in particular spatial
layout format.  Color features are often
easily obtained from the logo image. The color histogram is probably one of the
most popular gross representations of the foreground object in which the
precise spatial information is lost, so an extraction is generally impossible.
Since logo is designed with a setting of few color combinations, so color will
be ignored as far as the unique identity of a logo which is represented as an
intrinsic graphic pattern. On the other hand, the text in the logo is often modified
to add to its aesthetic appealing; its segmentation or the OCR processing may
not be easy and also unnecessary for the logo handled by a shape analyzer.
Similarly, if a logo contains texture patterns, the texture patterns can be
treated as a graphic pattern and handled with other parts together. Hence,
shape analysis of the logo is the main concern here.

 

3.   Literature survey

 

Previously there
are several contributions made by eminent people for logo matching and
recognition. Apostolos P. Psyllos, Christos-Nikolas E. Anagnostopoulos proposed
a SIFT-based enhancement matching scheme algorithm for vehicle logo recognition.
This algorithm is assessed on a set of nearly 1200 logo images that belong to
ten distinctive vehicle manufacturers. It is shown that the proposed enhanced
matching approach boosts the recognition rate in vehicle logos making it
suitable for real-time applications. In all the previous methods they provided
excellent results for image or logo matching and detection. But the multiple
descriptor methods are also used previously for the fake logo detection.But in
this paper we propose enhancement of the image resolution along with accuracy.
Smita A. Patil and T.B. Mohite Patil proposed a solution for logo recognition
based on context dependent similarity that directly incorporates the spatial
contact of local features. It is suitable to detect similarities and
differences between both nearer and duplicate logos by intensity matching. The
solution is proved to be highly effective and responds to the requirement of
logo detection and recognition in real world images. Ke Gao, Shouxun Linl,
Yongdong Zhang , Sheng Tang, Dongming Zhang collected a data set of 10,016
images from the web and TV shows. Ten popular logos are selected for testing such
as Starbucks, Coca–Cola, Nike, CNN, etc. These logos are either on products or
in cluttered background, and very different in position, orientation, and
sizes. Related images for each logo category vary from 30 to 200, and the rest are
non-related images also contained as interference.

 

4.    Objectives

 

The
multiple descriptors are scale invariant feature transform (SIFT), Speeded up
robust feature (SURF), histogram oriented gradient (HOG) and Gradient location
and orientation histogram (GLOH). By using this method we assure high
resolution and great accuracy.

 

5.   Methodology going to be
used

 

In this work we proposed
multiple descriptors method along with context dependent similarity for
enhancing the accuracy and resolution of the logo. This approach for logo recognition
and detection depends on the description of a context dependent similarity
(CDS) kernel. This method is used for descriptors and is not limited to any
derivable arrangement model. It is mainly deals with the spatial context of
local features and the scheme of recognition process. The data set is obtained
from google for sample testing. All these logos are in TIFF format and resized
to 150×150 pixels, converting images into binary image and into double.

 

 

 

Figure: Expected
Algorithm.

6.    Possible outcome /
Result

 

By using MATLAB tools this method is implemented. This
is a novel logo detection and localization approach based on a new class of
similarities referred to as context dependent. In several aspects the proposed
system strength resides. So many numbers of popular logos were collected for
the implementation of this method and formed as data set with the particular
name. We have chosen an image during execution process and then select another
logo which is need to compare pixel intensity values with another logo. Between
these logos we find more compatibility. Finally justifying the logo is original
one or not. The following table describes the percentage of accuracy got with
different process. And the graph shows precession and recall values.

 

Logo Matching with Different Descriptor Combination:

 

TABLE
1: THE COMPARISON SAMPLE RESULT

 

Process

Percentage Accuracy

Only SIFT

56.66%

Only HOG

76.66%

Only SURF

81.66%

SIFT with HOG

73.33%

SIFT with SURF

83.33%

SURF with HOG

85.00%

SIFT , HOG and SURF

91.60%

SIFT,HOG,SURF and GLOH

96.66%

Fig : The graph between precision Vs Recall

 

 

x

Hi!
I'm Johnny!

Would you like to get a custom essay? How about receiving a customized one?

Check it out