A novel LSTM-RNN decoding algorithm in CAPTCHA recognition.pdf

A novel LSTM-RNN decoding algorithm in CAPTCHA recognition.pdf

This preview shows page 1 - 2 out of 6 pages.

This paper is partially supported by Natural Science Foundation of AnHui Province under grant no. 1208085QF107 A novel LSTM-RNN decoding algorithm in CAPTCHA recognition Chen Rui, Yang Jing, Hu Rong-gui, Huang Shu-guang Department of network Electronic Engineering Institute Hefei, China e-mail: [email protected] Abstract LSTM-RNN has been succeeded in applying in offline handwritten recognition. The paper used two-dimensional LSTM-RNN to recognize text-based CAPTCHA. Aiming at the problem that traditional decoding algorithm cannot obtain satisfactory results. The paper proposed a novel decoding algorithm based on the multi-population genetic algorithm. The experimental results showed that the novel decoding algorithm for the LSTM-RNN can improve the recognition rate of merged-type CAPTCHA. Keywords-Network Security; CAPTCHA recognition; decoding algorithm; Recurrent Neural Network I. I NTRODUCTION Completely Automated Public Turning test to tell Computers and Humans Apart (CAPTCHA) provides a way for automatically distinguishing a human from a computer program, thereby avoiding abuse of the network resource by the computer program. CAPTCHA is a kind of network security mechanism based on hard artificial intelligence problems. Currently the research on CAPTCHA is focused on the design and recognition technology[1-3]. In this paper, the recognition technology of the text-based CAPTCHA is chiefly studied. Nowadays the researchers mainly use the technology of pattern recognition for CAPTCHA recognition[4] such as SVM, Neural Network, HMM and so on. These methods are all based on segmentation, however once character in the CAPTCHA touching or merged are encountered, the segmentation becomes difficult. A Segmentation-free strategy based on two dimension Long-short Term memory Recurrent neural network(2DRNN) [5] is used for merged-type CAPTCHA recognition without segmentation in this paper. Aiming at the problem that traditional decoding algorithm cannot obtain satisfactory results, A decoding algorithm based on the multi-population genetic algorithm is proposed in this paper, and we call it decoding algorithm based on GA for short. II. R ELATED W ORK The recognition technology of text-based CAPTCHA normally include these steps: Firstly at preprocess stage the noise and complex background in the CAPTCHA image are cleared, then machine learning algorithm is used for the character string recognition. The recognition technology can be divided into segmentation-based strategy and segmentation-free strategy. In the segmentation-based recognition systems, the boundary of the character is determined first, and then the CAPTCHA image is segmented into individual character image, finally the character image is recognized.
Image of page 1

Subscribe to view the full document.

Image of page 2

{[ snackBarMessage ]}

What students are saying

  • Left Quote Icon

    As a current student on this bumpy collegiate pathway, I stumbled upon Course Hero, where I can find study resources for nearly all my courses, get online help from tutors 24/7, and even share my old projects, papers, and lecture notes with other students.

    Student Picture

    Kiran Temple University Fox School of Business ‘17, Course Hero Intern

  • Left Quote Icon

    I cannot even describe how much Course Hero helped me this summer. It’s truly become something I can always rely on and help me. In the end, I was not only able to survive summer classes, but I was able to thrive thanks to Course Hero.

    Student Picture

    Dana University of Pennsylvania ‘17, Course Hero Intern

  • Left Quote Icon

    The ability to access any university’s resources through Course Hero proved invaluable in my case. I was behind on Tulane coursework and actually used UCLA’s materials to help me move forward and get everything together on time.

    Student Picture

    Jill Tulane University ‘16, Course Hero Intern