为克服传统二维文本验证码的局限性，文本验证码演变出一些新的形式，其中包括3D文本验证码。针对目前网站上使用的一种3D文本验证码，文中提出有效的破解方法。利用图片中像素密度首先从验证码图片中提取字符边界；再通过图像中背景纹理梯度方向基本不变的特征，从验证码图片中提取字符背景，从而间接得到字符表面；然后根据字符信息的表现特征，设计字符分割算法，以得到验证码图片的单个字符；最终采用OCR识别软件———ABBYY进行字符识别。实验结果表明，提出的破解算法在实验数据集上取得了较好的破解效果。充分利用了验证码系统的规律，通过提取图片背景间接得到字符前景。与基于直接提取字符表面的破解算法相比较，前者具有更好的适用性。%In order to overcome the traditional limitations of two-dimensional text CAPTCHAs,there comes many new forms,including 3D text CAPTCHAs. Aiming at authentication code of 3D text using on site currently,an effective crack method is put forward. Using fea-tures in pixel space,such as pixel density and gradient direction of diagonal lines,the character boundaries and surface from image is ex-tracted. Secondly,character segmentation algorithm is designed to get single character according to the information of text. Finally,using OCR for identification. The experiment shows that the breaking algorithm proposed achieves a good result on experimental data set. Mak-ing full use of the rule of CAPTCHA scheme,characters from images are obtained indirectly. Compared with the algorithm which extracts characters directly,it has a better applicability.
展开▼
