Study On The Recognition Of Validation Code Based On Neural Network

With the rapid development of network technology, Internet in providing to express people’s life at the same time, security can not be overlooked, the network verification code as an effective means to protect the network security, has been widely used. Strengthen verification code identification technology, for overcoming the loopholes in the verification code, increasing the resistance of malicious attacks, and protect the network security has important significance.The identification of the verification code needs a complicated procedure and procedure, and every step and step will affect the verification of the verification code. Verification code technology includes image preprocessing, character segmentation and character recognition and matching technology.. There is a difference in the use of code knowledge under different conditions, so it has important practical value for the research of verification code.In this paper, the legend of the blood as an example, for different verification code analysis of its unsafe performance and the way of the crack. And it also describes the different kinds of recognition algorithms, which makes the research of theoretical and practical significance.. In this paper, we focus on the BP neural network and convolutional neural network, and make a detailed deduction for them.. The problem based on the legend of Mir verification code exists, for to put forward application of vertical projection algorithm to achieve the segmentation, thus avoiding recognition of merged characters and the superposition of defect verification, using a convolutional neural network approach for a single character identification, recognition rate more than 50%.In this paper, the verification code of the research is relatively simple.. The algorithm mentioned in this paper is the arithmetic of the characters and the characters of the alphabet.. With the development of the application of the verification code technology, the requirements for verification code are higher and more complex, and even the identification of Chinese characters, etc.. Verification code recognition technology to further research and development.