基于属性网格计算的验证码识别研究

Study on Recognition of CAPTCHA Based on Attribute Grid Computing

文中旨在基于对属性网格计算的研究基础上,提出一种新的验证码识别研究方式,并通过实验验证该方法的可行性。方法如下:在验证码识别之前,需要先进行预处理,其中包括:图像的灰度化、二值化、去噪、分割以及归一化。在识别部分中,会用到属性网格计算器,这是一种基于定性映射的新型计算器。以对象的不同属性特征为维度,建立属性网格,并根据属性网格做出计算。首先对标准字符建立属性网格,然后在每次识别字符时,对识别字符的属性进行网格计算,并得出结果。实验结果成功率较高。该结果证明了该方法的可行性,同时表明了识别成功率与不同属性的权值分配有关。

A new method is expected for CAPTCHA recognition in this paper, which is based on research on attribute grid computing, and the experiments show this method is feasible. There are several steps in the preprocessing of CAPTCHA, which include transforming im- age to grayscale image,image binarization, image denoising,image segmentation and normalization. In recognition part, use attribute grid computer, which is a kind of new computer based on qualitative mapping, build a attribute grid for object with its attributes and compute according to it. In this paper, build a attribute grid for standard characters, then, once need to recognize character, this grid can be used for computing and getting the results. The success rates of experiments are high, which verified the feasibility of this method, and also veified that the success rates Of recognition are related with the allocation of the weights among different attributes.