See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/224310554

Advanced Collage CAPTCHA

Conference Paper · May 2008

DOI: 10.1109/ITNG.2008.12 · Source: IEEE Xplore

CITATIONS READS
20 55

2 authors, including:

Mohammad Shirali-Shahreza
Sharif University of Technology
97 PUBLICATIONS 1,105 CITATIONS

SEE PROFILE

All content following this page was uploaded by Mohammad Shirali-Shahreza on 24 July 2014.

The user has requested enhancement of the downloaded file.

 Fifth International Conference on Information Technology: New Generations

Advanced Collage CAPTCHA

Mohammad Shirali-Shahreza Sajad Shirali-Shahreza

Computer Science Dept., Sharif University Computer Engineering Dept., Sharif University
[email protected] [email protected]

Abstract One of the Non-OCR-Based CAPTCHA methods is

Collage CAPTCHA [2] which is designed by authors.
Some Internet websites for providing services to In this method image of some shapes is shown with
their customers, ask them to register in the website. distortion and the user is asked to click on a certain
Unfortunately hackers write programs to make object. In this paper we proposed a method for
automatic false enrolments which waste the website’s increasing the resistance of this method to hackers’
resources. To solve this problem, systems known as attacks. We will discuss our method and its
CAPTCHA (Completely Automated Public Turing test experimental results in third section. In the last section
to tell Computers and Humans Apart) have been we make the final conclusion.
introduced to distinguish between human user and
computer program. One of the CAPTCHA methods is 2. Related works
Collage CAPTCHA. In this method some shapes are
shown with distortion and the user is asked to choose a CAPTCHA systems are usually used to prevent
specific object. fake registration by computer programs in websites,
In this paper we increase the resistance of this but can be applied to other cases such as phishing
method to attacks. For this purpose, we also show attacks [3]. CAPTCHA methods can be divided into
some other objects on the right of the screen. Some of two groups: OCR-based and Non-OCR-Based.
these objects are same as the previous objects, but In OCR-based methods, the image of a word with
have the different shapes. Now we ask the user to distortion and various pictorial effects is shown to the
choose a specific object and also its same object on the user and he/she asked to type that word. Due to
right of the screen. The user will be passed the test if presence of various pictorial effects, the computer will
he chooses the two similar objects correctly. encounter problems in the recognition of these words
In this method because the computer program and only a human user can recognize the word.
should also recognize the similar object, the possibility Examples of these methods include Gimpy [1] and
of passing the test by computer is more reduced. Persian/Arabic Baffle Text [4]. But these methods
usually result in dissatisfaction of users.
1. Introduction In contrast, we can point to Non-OCR-based
methods which are easier to work with than OCR-
CAPTCHA (Completely Automatic Public Turing based ones. These methods mainly based on the
Test to Tell Computer and Human Apart) systems are features of multimedia systems such as pictures and
used to distinguish between human users and computer sounds and usually using methods like small puzzle
programs automatically. These systems are based on games. Examples of these methods include PIX [1] and
Artificial Intelligence (AI) topics. They are similar to Collage CAPTCHA [2].
Turing test, but they differ in that the judge is a
computer [1]. The goal of these systems is to ask 3. The suggested method
questions which human users can easily answer, while
current computers programs cannot. CAPTCHA In this paper, we improved the resistance of Collage
methods can be generally divided into two groups: CAPTCHA method [2]. In this section, after
OCR-based and Non-OCR-Based. In the next section, describing the original Collage CAPTCHA method,
we will discuss these systems in more details. we will describe the suggested algorithm.

978-0-7695-3099-4/08 $25.00 © 2008 IEEE 1235

1234
DOI 10.1109/ITNG.2008.12
 3.1. Collage CAPTCHA method Before asking the user to click on the picture of
goal object, the program selects 6 objects which are
Collage CAPTCHA is a method for distinction different from previous objects. However the objects
between human users and computer programs through can be the same, but has different shape. All pictures
recognition and finding a picture of an object among are rotated a little. The sixth object is selected as the
some objects. The structure of the suggested method is same to goal object, but with different shape. Finally
as follows: the program asks the user to click on the picture of
At first a bank of the pictures of objects is prepared. goal object and its similar object on the right side of
These pictures include objects, animals, different the screen. If the user chooses both of pictures
persons, flags of the countries, etc. The concerned correctly and clicks on them, he/she will be allowed to
program chooses some of these pictures (e.g. 6 carry out the concerned operations.
pictures) randomly. Each of these pictures is rotated a
little. Then all the pictures are shown on the screen. 4. Conclusion
The pictures are putted in random places but the
pictures don't overlap. In this project we improve the Collage CAPTCHA
Computer chooses one of the shown pictures as a method by presenting a method for increasing the
goal and asks the user to click on the picture of that resistance of it.
object. If the user clicks on the goal picture, we can Although we can increase the rate of its difficulty in
conclude that a human user has done the click and the order to improve its resistance against the attacks by
user is not a computer program because a computer applying other effects such as increasing the objects
program should do two operations successfully in present in the screen and decreasing the distance
order to respond appropriately: First it must recognize between the objects, this way the test will become
the wanted object and has knowledge of its shape. more difficult even for a human user in addition to
Then it must find the place of the object on the screen. computer programs.
We can also implement this method on other
3.2. The improved method devices such as mobile phone, PDA (Personal Digital
Assistant), and the devices which have touch screens,
Our method is aimed to increase the resistance of because no keyboard is needed in this method and also
Collage CAPTCHA method to hackers’ attacks. The there is no need to heavy processing. Besides its
structure of this method is as follows: advantage of covering every age group, even children,
After choosing one of the shown pictures as a goal we can also recommend it for disabled people.
and before asking the user to click on the picture of We can make this method specialized for specific
that object, the program selects some other object and websites for example we can use the specialized sport
put them on the right side of the screen. These pictures pictures (such as soccer players pictures) for sport
are also rotated a little. One of these pictures is the news websites.
same as the goal picture, but has different shape.
Now the computer program asks the user to choose 5. References
that object and also the same object on the right screen.
If the user selects both of them correctly, we can guess [1] L. von Ahn, M. Blum, and J. Langford, “Telling Humans
that the user is a human being not a computer program. and Computers Apart Automatically,” Communications of
In the original Collage CAPTCHA method the the ACM, vol. 47, no. 2, February 2004, pp. 57-60.
computer requires two abilities to pass the test: [2] M. Shirali-Shahreza and S. Shirali-Shahreza, “Collage
1. To find out the shape of the concerned object. CAPTCHA,” Proceedings of the 20th IEEE International
2. To find the concerned object on the screen. Symposium Signal Processing and Application (ISSPA
2007), Sharjah, United Arab Emirates, February 12-15, 2007.
Considering the insufficient ability of the computers [3] R. Dhamija and J. D. Tygar, “Phish and HIPs: Human
to carry out these tasks, only a human user can interactive proofs to detect phishing attacks,” Proceedings of
recognize and choose the concerned object. Second International Workshop on Human Interactive
But now in addition to above tasks, computer Proofs: (HIP 2005), PA, USA, May 2005, pp. 127-141.
should find two objects which are similar to each [4] M.H. Shirali-Shahreza and M. Shirali-Shahreza,
other. Therefore the resistance of method is increased. “Persian/Arabic Baffletext CAPTCHA,” Journal of
This method is implemented by Java programming Universal Computer Science, vol. 12, no. 12, Dec. 2006, pp.
language. The implementation is similar to original 1783-1796.
Collage CAPTCHA method with some differences.

1235
1236

View publication stats