In this modern era of online world, one requires to have an online account. It may be a social, transactional, online shopping or an e-mail account, with a heap of so many accounts and the burden of memorizing the difficult passwords, user tends to compromise on the password, thereby weakening the security aspect. Most of the passwords are prone to shoulder surfing attack. In this paper we propose a novel Graphical authentication system named Confusys. Cofusys system takes care of the shoulder surfing assault, as it deals with information from different images. The befog module does not provide any information to the attacker; above all it confuses the attacker more .The befog module differs for every user as it is purely based on the click point from the user. The slightest change in the click point from the user results in a massive change of data. A prototype of this system is implemented to check for various parameters of this system and the results proved the chances of breaking into the system is very less within the range of 7.63183 e^-15 to 1.38936 e^-17, and also provides password entropy between 2⁵⁸ to 2⁶⁷.