| With the development of the computer hardware and software, for the networksolves threatens which come from the automatic program in the Internet, CAPTCHAtechnology appears. The appearance of CAPTCHA distinguishes the differencebetween human and automatic program, and protects the network from the destructiveoperation. The forward research on CAPTCHA recognition technology can not onlyimprove the technology which is developed by now, but also promote the evolution inartificial intelligent.GPU had been changed a lot in the recent years, unlike it could be only used in theimage process area in the early years, now it has been developed its ability to thegeneral propose computation. It has been used in the applications of MRI fast analysis,the simulation of the molecule motion and hedging model in financial area.In this thesis, the author combines the research of CAPTCHA technology and thehigh performance computation platform in GPU, with the help of the GPU to improvethe traditional HTM algorithm. For proving the idea, the author searches the relativetheory deeply and does some practical operation.In the beginning of the thesis, it introduces the derivation and development statuesof CAPTCHA and GPU, and it shows some relative work which were done by insideand outside of our country. The meaning and application of this research is alsoindicated in the first chapter. Then, for the easy understanding of readers, it shows thealgorithm will be used latterly by classify them into a few classes. Based on theprevious introduction, author puts up a new algorithm which is called P-HTM algorithm.Author shows the algorithm in detail thinking and streaming. Then, Achieving a systemwhich is built based on the algorithm, it will include the whole thinking of the algorithmand have the extend implement. At last, a rigid test will take on the system and thealgorithm, to make sure the system has been achieved all the functions and thealgorithm is efficiency. |
PDF Full Text Request