Character identification method and apparatus

Abstract

Embodiments of the present invention provide a character identification method and apparatus. The method comprises: generating a to-be-identified high-pixel image by performing enlargement processing on a to-be-identified image; performing binarization processing on the to-be-identified high-pixel image, and filtering out a disturbing factor in the to-be-identified high-pixel image; acquiring a picture containing a to-be-identified character from a binarized image by adopting a projection histogram method; and inputting the acquired picture containing the to-be-identified character into a multi-layer artificial neural network model, and acquiring the to-be-identified character output by the multi-layer artificial neural network model. By adopting the technical scheme of the present invention, binarization processing is performed on the to-be-identified image, so that disturbing factors such as a background image can be effectively filtered out, thereby ensuring accuracy of a final recognition result. The Binarized image is segmented by adopting the projection histogram method, so that it is ensured that each picture containing the to-be-identified character that is obtained through segmentation contains a complete character, and accuracy of the recognition result is higher.

Description

technical field [0001] Embodiments of the present invention relate to the field of image processing, and in particular, to a character recognition method and device. Background technique [0002] A captcha is a computer program that distinguishes whether a user is a computer or a human. Verification code technology can effectively prevent malicious cracking of passwords and frequent automatic submission of certain tasks through a mandatory user interaction process. [0003] With the development of computer technology, the verification code can also be identified by using graphic image processing technology. Specifically, the picture where the verification code is located is segmented, and the divided picture is identified to determine the verification code included in the picture. Among them, the following two methods are usually used to segment the picture where the verification code is located. The first segmentation method is based on the preset segmentation width. The ...

AI Technical Summary

Problems solved by technology

[0005] Embodiments of the present invention provide a character recognition method and device to solve the problem of large errors in character recognition results in the current process of recognizing characters contained in verification codes

Images