|
The Effects of Partial Occlusion in Image-based CAPTCHAs on Pass Rates |
|---|---|
| รหัสดีโอไอ | |
| Creator | ฐปกรณ์ กฤษฎารักษ์ |
| Title | The Effects of Partial Occlusion in Image-based CAPTCHAs on Pass Rates |
| Contributor | ณัฐธนนท์ หงส์วริทธิ์ธร |
| Publisher | Faculty of Information Science and Technology, Mahanakorn University of Technology |
| Publication Year | 2555 |
| Journal Title | Journal of Information Science and Technology |
| Journal Vol. | 3 |
| Journal No. | 1 |
| Page no. | 23-32 |
| Keyword | CAPTCHA, imaged-based CAPTCHA, Gestalt Theory |
| URL Website | https://tci-thaijo.org/index.php/JIST |
| Website title | Journal of Information Science and Technology |
| ISSN | 2651-1053 |
| Abstract | Several attempts in research of CAPTCHAs are to design and develop different kinds of CAPTCHAs in order to make them more deployable for computer system protection from internet bots. Text-based CAPTCHAs are commonly used in practice. They have been increasingly more distorted and camouflaged due to the increase of advanced developed bots. However, the increasing distortion of CAPTCHA text makes it difficult for human to decode and then take more time to get into systems. This research was to proposed new image-based CAPTCHAs which was developed base upon human visual perception so that the CAPTCHA is more usable to humans, but hard to decipher for auto-bots. The proposed image-based CAPTHCAs were generated based upon three factors to make the CAPTCHAs incomplete. Those factors are the two partial occlusion methods (constant and random occlusion of images), the two levels of image occlusion (30% and 45%), and the three different number of small frame as interference in an image (no frame, two frames and three frames). The proposed CAPTCHAs were evaluated by accuracy rate (pass rate) together with user satisfaction. A within-Subject design experiment was run with 40 participants. The results of the experiment indicated that the participants were satisfied with the image-based CAPTCHAs more than text-based CAPTCHAs at the 95% significant level. And importantly, the accuracy rate of image -based CAPTHCAs was significantly higher than text-based. The response time will be different depending on the completion of image CAPTCHAs. The most usable image-based CAPTCHAs should be made incomplete by constant occlusion of 30% of original picture, and can be interfered with zero or two small frames. |
