Abstract—A CAPTCHA is a type of challenge-response test used to determine whether the user is human or not. This user identification procedure has received many criticisms from people who feel the distorted words are illegible even for users with no disabilities. In order to prevent website owner from suffering attacks and simultaneously maintain the readability of generated CATCHA, this paper proposed an improvement of image-based CAPTCHA which embeds versatile characters in the images for distinguishing human and computer. The improvement selects the locations for characters smartly by a salient detection of the used image. The greater salient location is picked to embed the versatile characters. By calculating the integral image of the salient image, we can decide the appropriateness of the selected location quickly. The proposed method makes the characters indiscernible by automated pattern recognition technologies like optical characters recognition while human can easily distinguish the location of the embedded characters. In experiments, 10 users were invited to test the system and the success rate is about 95.6%. Compare the average logging time with previous studies, the proposed method is also faster.
 

Index Terms—CAPTCHA, internet verification code, image analysis, salient detection

Cite: Chen-Chiung Hsieh, Yu-Chen Lin, Kai-Miao Cha, and Chao-Wei Huang, "Location Selection for Versatile Characters of Image Based CAPTCHA," International Journal of Signal Processing Systems, Vol. 4, No. 3, pp. 239-244, June 2016. doi: 10.18178/ijsps.4.3.239-244