A CAPTCHA Recognition Algorithm Based on Holistic Verification
Shuguang Huang, Liang Zhang, Wang Peng-Po, Hongwei Han · 2011
CAPTCHA is a new kind of network security mechanism. Studying the recognition of CAPTCHA can help to discover its hidden defects and thus make it more secure. For closely-connected CAPTCHAs that can hardly be recognized by methods of state of art, this paper proposed a new recognition algorithm based on holistic verification. During the process of this algorithm, Recurrent Neural Network (RNN) was first used to recognize unknown CAPTCHAs. Then, recognition results were verified by SVM rejection, synthetic data generation and Extreme Learning Machine (ELM). Experiments results show that this algorithm can not only recognize closely-connected CAPTCHAs but also effectively boost the recognition rate of RNN.