基于摄像头图像的汉信码识别系统及相关加密算法

发布时间：2018-05-10 10:24

本文选题：汉信码 + 预处理　；参考：《吉林大学》2014年硕士论文

【摘要】：作为具有自主知识产权的免费二维码体系，，汉信码所具有的优势主要体现在其超强的汉字处理能力.有鉴于此，汉信码在日常实践中的使用日益广泛，对汉信码的研究也逐渐成为热点.汉信码在增值税发票上取代普通密文用来存储发票信息，除票面上的数字信息外，对货物名称、规格型号、单位都能进行存储，从而有效的避免了随意开发票、伪造发票等非法行为.目前，利用普通扫描仪采集发票票面信息，然后应用解码算法还原汉信码票面信息的技术已经比较完善，2012年后此项技术已经逐渐投入使用，其识别率也由最开始的85%提升到了现在的98%. 但是，设备价位、设备噪声和图像采集速度等问题严重限制了汉信码技术的应用范围，影响了汉信码本身的应用前景.因此，最近市场开始出现用数码摄像头作为拍照方式采集汉信码图像的设备，这虽然在一定程度上解决了上述问题，但设备上的部分功能缺失也同时大幅度提高了汉信码的解码难度，从而对我们的解码算法提出了更高的要求.这就需要我们对处理扫描仪图像的算法进行大规模的改进，这些改进主要集中在图像采集、预处理等几个方面.例如，在图像采集时遇到的红色印章过滤、亮度不均匀等问题上，我们就有针对性的采用了图像拼接技术以及白平衡算法对其进行处理，也得到了较好的效果，几乎没有影响到识别算法的识别率. 针对汉信码识别算法中存在的部分问题，我们重新做了调整，其中包括发票码图区域的倾斜角度的求取、畸变校正以及插值函数的选取等，并且提出了改进方法.具体说来，我们应用随机Hough变换计算汉信码的倾斜角度；在插值函数选取方面通过实验对比发现双三次插值比双线性插值更加适用于我们采集的汉信码图像，不仅提高了精度，而且在时间复杂度方面也未出现明显的差异；另外，利用二次差分、积分方法我们在码图分离方面都取得了很好的效果. 2013年火车票QR码被破译的新闻传出，我们对汉信码的安全问题同样表示担忧，因此在汉信码信息安全方面值得关注，我们用嵌入数字水印的方法来实现汉信码的加密，对编码信息直接进行加密能够有效的避免常规识别软件的成功解码，应用Arnold置乱算法对编码信息加密，其变换次数具有一定的周期性，所以选择恰当的变换次数对加密的有效性很重要.
[Abstract]:As a free QR code system with independent intellectual property rights, Hanxin Code has the advantages of its strong Chinese character processing ability. In view of this, the application of Han letter code in daily practice is becoming more and more extensive, and the research of Chinese letter code has gradually become a hot spot. Han Xin code replaces ordinary ciphertext on VAT invoices to store invoice information. In addition to the digital information on the face, the names, specifications, models and units of the goods can be stored, thus effectively avoiding the arbitrary invoicing. Illegal acts such as forgery of invoices. At present, the technology of collecting invoice face information by ordinary scanner and applying decoding algorithm to restore ticket face information of Hanshin code has been quite perfect. After 2012, this technology has been gradually put into use. The recognition rate also rose from 85% at the beginning to 98% now. However, the problems of equipment price, equipment noise and image acquisition speed seriously limit the application scope of Hanshin code technology, and affect the application prospect of Hanshin code itself. As a result, recently, the market has begun to use digital cameras as a means of capturing Hanshin images, although to some extent this has solved the above problems. However, the lack of some functions on the equipment also greatly increases the difficulty of decoding the Hanshin code, which puts forward higher requirements for our decoding algorithm. Therefore, we need to improve the algorithm of scanner image processing on a large scale. These improvements mainly focus on image acquisition, preprocessing and other aspects. For example, in the image collection encountered in the red seal filtering, brightness uneven and other problems, we have targeted the use of image stitching technology and white balance algorithm to deal with it, but also got a better effect. There is little effect on the recognition rate of the recognition algorithm. In order to solve some problems existing in the algorithm, we re-adjust it, including the calculation of tilt angle of invoice code area, the correction of distortion and the selection of interpolation function, and the improved method is put forward. In particular, we use the random Hough transform to calculate the skew angle of the Chinese signal code, and in the aspect of the interpolation function selection, we find that the bicubic interpolation is more suitable than the bilinear interpolation for the Chinese signal code images collected by us, which not only improves the precision, but also improves the accuracy of the interpolation function. Moreover, there is no obvious difference in time complexity. In addition, we have achieved good results in code graph separation by using quadratic difference and integral method. The news that QR code of train ticket was deciphered in 2013 came out, so we also expressed concern about the security of Hanshin code, so we should pay attention to the security of Hanshin code information. We use the method of embedding digital watermark to realize the encryption of Hanshin code. The direct encryption of the coded information can effectively avoid the successful decoding of the conventional recognition software. The Arnold scrambling algorithm is used to encrypt the coded information, and its conversion times have a certain periodicity. Therefore, it is very important to choose the appropriate number of transformations for the effectiveness of encryption.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP391.44

【参考文献】