Propose Image Captcha System

Abstract

Captcha has been designed to be easy for humans and hard on the machines. Captcha are used by many websites to determine
human users from indecent programs. This paper will propose two types for generating captcha these are; text and image captcha. In the first type use (6*6) captcha table to store alphabetic characters 'A'..'Z' and '0'..'9' and use random function to generate six numbers to represent the row and colum numbers which used as index to retrieve the character from the captcha table to get the final captcha. In the second types a sample of 100 images from PayPal.com Human Interactive Proofs (HIP) are used. In order to recognize these images three steps for recognize the characters (pre-processing, segmentation and classification) are adopted. The modification was in the preprocessing step where proposing to use Gabor filter to remove the noise in the images which make the recognition
accuracy 100% which is good results when compared with the obtained accuracy in reference [1] where the accuracy value in the range (80% .. 98%) for different scheme used in this work.