低质量DM码图像识别技术研究

发布时间：2019-04-20 15:02

【摘要】：在物联网领域,二维条码是一种非常重要的信息识读技术。作为一种新型信息传输媒介,二维条码具有信息容量大、准确性和实时性等性能。目前,二维条码识别技术已经引起了工业界和学术界广泛的研究兴趣,国内外公司和科研单位竞相投入大量资源进行技术研发。对于二维条码技术研究,编码技术已经形成了比较成熟的国际标准。其研究难点主要在于二维条码解码。特别是部分待解码的条码图像,由于受拍摄角度、光照、打印质量等,其图像质量非常低。因此,如何从这些图像中有效恢复出编码结构并识别是一个急需解决的问题,这正是论文的研究重点。围绕二维DM码图像的解码问题,论文主要研究工作如下：1.二维条码图像预处理。针对二维条码图像进行增强处理,提高对比度。2.改进的L形定位算法：针对点阵DM条码图像L形特征不明显问题,提出结合Max-Min算法和Canny边缘检测算法,较好地增强了点阵码的L形特征。3.局部特征自适应的结构提取方法：针对非规则变形导致条码模块分布不均匀,失真比较严重导致较难准确地恢复出条码结构。针对该问题,论文提出在全局仿射变换基础上,通过边缘特征对模块划分进行自适应调整,从而使得划分结果能够有效地反映出条码黑白分布特征,重建出原有条码结构。实验表明,本文对行列分布不一致的机器点阵码、高度不一致的打印码等,都能够得到非常好的解码效果。4.二维条码识别核心组件实现。在上述研究基础上,论文实现了一个二维条码识别核心组件,并应用于社保信息采集以及工业码识别。该系统包括图像预处理、条码结构恢复、二维条码识别模块。并通过大量的实际二维条码进行实验分析,验证了算法的有效性和实用性。
[Abstract]:In the field of Internet of things, two-dimensional bar code is a very important information recognition technology. As a new type of information transmission medium, two-dimensional bar code has large information capacity, accuracy and real-time performance. At present, two-dimensional bar code recognition technology has aroused a wide range of research interests in industry and academia, and domestic and foreign companies and scientific research institutions have invested a great deal of resources in technology research and development. For the research of two-dimensional bar code technology, coding technology has become a mature international standard. The research difficulty lies in the decoding of two-dimensional barcode. Especially for some bar code images to be decoded, the image quality is very low because of the photographed angle, illumination, print quality and so on. Therefore, how to effectively recover the coding structure from these images and identify it is an urgent problem, which is the focus of this paper. Focusing on the decoding of two-dimensional DMcode images, the main research work in this thesis is as follows: 1. Two-dimensional bar code image preprocessing. Enhancement of two-dimensional bar code image to improve contrast. 2. Improved L-shape location algorithm: aiming at the problem that the L-shape feature of dot-matrix DM bar code image is not obvious, a combination of Max-Min algorithm and Canny edge detection algorithm is proposed to enhance the L-shape feature of dot-matrix code. 3. The local feature adaptive structure extraction method: aiming at irregular deformation, the distribution of bar code module is not uniform, and the distortion makes it difficult to recover the bar code structure accurately. In order to solve this problem, based on the global affine transformation, this paper proposes to adjust the module partition adaptively through edge feature, so that the partition result can reflect the black-and-white distribution of barcode effectively and reconstruct the original barcode structure. The experimental results show that this paper can get very good decoding results for the machine dot codes with inconsistent column and column distribution, and the highly inconsistent printing codes, etc. Realization of two-dimensional bar code recognition core component. On the basis of the above research, a two-dimensional bar code recognition core component is implemented, which is applied to the collection of social security information and the identification of industrial codes. The system includes image preprocessing, barcode structure restoration and two-dimensional bar code recognition module. The validity and practicability of the algorithm are verified by experimental analysis of a large number of actual two-dimensional bar codes.
【学位授予单位】：浙江工业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP391.41

【相似文献】