复杂环境下刀具DPM码识别技术研究

发布时间：2018-04-20 15:24

本文选题：刀具 + 直接标识码识读　；参考：《西北工业大学》2014年博士论文

【摘要】：二维条码作为一种新兴的信息识别技术，在产品销售、广告、互联网、移动通信、电子商务等领域具有广泛而成熟的应用。工业生产中，由于二维条码具有占用面积小、信息量大、可靠性高、纠错能力强及保密性好等多种优点，已成为直接零件标识（Direct Part Marking，DPM）技术的首选，其生成的二维条码作为机械零件、刀具等生产要素生命周期中的唯一永久标识，称为直接标识码或DPM码，可以实现产品信息的实时追踪和管理。由于金属刀具本身特有的不利识别因素以及流通中不可避免的污染因素，利用传统条码识读技术直接对标刻在刀具上的DPM码进行识别具有一定难度。因此，迫切需要进行面向刀具表面二维条码的识读方法研究。本文以生产车间中的刀具DPM码为对象，针对复杂环境下刀具条码易受到光照不均、曲面畸变、锈蚀磨损等因素影响，，传统技术无法准确识读的问题，提出了一套适合复杂环境的刀具DPM码识读方案，力图实现车间流通环境下刀具DPM码的准确识别与信息读取。论文的主要研究内容和成果如下：（1）提出了一种适用于刀具表面光照不均DPM码的识读方案。从光照不均问题产生的原因入手，针对条码图像的低对比度问题，利用人工蜂群算法自适应选取非完全Beta函数的参数，实现了DPM码图像的对比度自适应增强；针对刀具表面条码数据区域易被高光覆盖导致无法识读的问题，在分析其产生原因的基础上，结合优化设计的光源设备，提出基于随机共振原理的高光源头抑制算法。实验结果表明，通过从源头上抑制高光的产生，可以较好地解决刀具表面的光照不均问题。（2）提出了两种适用于产生曲面畸变的柱形刀具表面DPM码的识读方案。第一种方案结合相机自标定和透视投影原理，在考虑镜头非线性畸变的条件下建立柱面DPM码的畸变校正模型，实现柱面DPM码图像的畸变校正。第二种方案在图像拼接算法的基础上，结合配套的刀具条码序列图像采集设备，实现DPM码的完整拼接。由于小缝隙内的DPM码区域可近似为均匀无畸变，因此拼接出的DPM码可抵消柱面大曲率畸变。实验结果表明所提出算法对于刀具条码的柱面畸变具有良好的抑制效果。（3）提出了一种基于多信息决策级融合的刀具分类识别方法，利用刀具本身的重量、轮廓，以及残余条码纹理等特征，分别建立识别模型，并通过融合多个特征识别模型的综合决策完成刀具产品的确定和对应条码信息的恢复。实验证明所提出算法较好地解决了DPM码由于油污、磨损、锈蚀等原因导致失效、刀具追踪断路的问题，具有良好的实用性。（4）在上述研究成果基础上，开发出一套面向车间刀具DPM码的专用识读系统。分别对该系统的硬件装置与软件系统的设计与实现进行介绍，并以车间常用刀具为例介绍了该系统的具体应用。本文的研究成果将有助于DPM技术的深化和应用。初步的工业生产现场应用试验表明，论文所提出的相关原理和方法对于提高复杂环境下的刀具DPM码识别效果具有一定的实用价值。
[Abstract]:As a new kind of information recognition technology , two - dimensional bar code has been widely used in the fields of product sales , advertisement , internet , mobile communication , electronic commerce , etc . It has become the first choice for direct part marking ( DPM ) technology .

In this paper , based on the tool DPM code in the production workshop , aiming at the problem that the tool bar code is susceptible to uneven illumination , surface distortion , corrosion wear and other factors in the complex environment , the traditional technology cannot accurately read the problem , and a set of tool DPM code reading scheme suitable for complex environment is put forward , which aims to realize the accurate identification and information reading of the tool DPM code under the workshop flow environment . The main research contents and the results are as follows :

( 1 ) A reading scheme is proposed which is suitable for the non - uniform DPM code on the surface of the tool . The self - adaptive enhancement of the contrast of the DPM code image is realized by using the artificial swarm algorithm to adaptively select the parameters of the non - perfect Beta function based on the reason of the problem of uneven illumination .
A high light source head restraint algorithm based on the principle of stochastic resonance is put forward based on the analysis of the causes of the surface bar code data of the cutter . Based on the analysis of the reasons , the high light source head restraint algorithm based on the principle of stochastic resonance is put forward . The experimental results show that the problem of uneven illumination of the surface of the tool can be better solved by suppressing the generation of high light from the source .

( 2 ) Two kinds of reading schemes are proposed for the surface DPM code of cylindrical cutter . The first scheme combines the camera self - calibration and perspective projection principle , establishes the distortion correction model of the cylindrical DPM code under the condition of nonlinear distortion of the lens , and realizes the complete splicing of the DPM code . Because the DPM code region in the small gap can be approximately uniform and no distortion , the DPM code can counteract the large curvature distortion of the cylindrical surface . The experimental results show that the proposed algorithm has good effect on the cylindrical distortion of the tool bar code .

( 3 ) A kind of tool classification and recognition method based on multi - information decision - level fusion is proposed . The recognition model is established by using the weight , contour and residual bar code texture of the tool itself . The determination of the tool product and the restoration of the corresponding bar code information are accomplished by combining multiple feature recognition models . The experimental results show that the proposed algorithm can solve the problem of failure of the DPM code due to oil pollution , wear and rust , and has good practicability .

( 4 ) On the basis of the above research results , a special reading system for the DPM code of workshop tools is developed . The design and implementation of the hardware device and software system of the system are introduced respectively , and the specific application of the system is introduced in the example of common tools in the workshop .

The results of this paper will contribute to the deepening and application of DPM technology . The preliminary industrial application tests show that the relative principles and methods proposed in this paper have some practical value to improve the efficiency of DPM code recognition under complex environment .

【学位授予单位】：西北工业大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TP391.44

【参考文献】