基于随机抽样一致性算法的PCB板定位方法研究  

Research on PCB board positioning method based on random sample consensus algorithm

作　　者：黄宝山[1] 邹浙湘[1] 温开元 吴光德 Huang baoshan;zou zhexiang;wen kaiyuan;wu guangde(zhuhai college, Belting university of technology, zhuhai, guangdong 519085)

机构地区：[1]北京理工大学珠海学院

出　　处：《自动化与仪器仪表》2018年第6期10-13,共4页Automation & Instrumentation

基　　金：2017年广东省自然科学基金(2017A030313291);2017年度广东大学生科技创新培育专项资金项目(PDJH2017b0924);2016年广东省高校省级重点平台和重大科研项目(2016KQNCX205)

摘　　要：随着对PCB板生产效率与质量的要求越来越高,针对目前PCB切割分离的分板机结构误差所导致的误切、过切的问题,文中提出了一种基于随机抽样一致性算法的PCB板定位方法。该方法首先在获取PCB板边缘图像后,采用自适应高斯平滑滤波对图像噪声区域进行平滑处理,降低图像模糊度。然后通过形态学运算和Sobel-Zernik算子提取PCB板轮廓样本点,接着在深入分析目标模型固有特性的基础上对样本点进行筛选,并采用分层抽样法分析样本点残差,经多次迭代之后得到最佳模型。文中最后将该算法与归一化的最小二乘方法、二次多项式拟合的边缘拟合法进行对比实验验证,通过在MATLAB中给图像加入高斯噪声和椒盐噪声以模拟不同环境下的图像,并分别使用归一化的最小二乘方法、二次多项式拟合的边缘拟合法与文中所提出的方法对图像进行Mark点识别。结果表明,文中所提出的基于随机抽样一致性算法的PCB板定位方法具有更高的稳定性与准确性,且鲁棒性好。As the higher requirements for the production efficiency and quality of PCB board is required. In order to solve the problem of miss cut and over cut caused by the structural error of PCB cutting and separating machine, a PCB board positioning method based on random sample consensus is proposed in this paper. Firstly, the edge image of the PCB is obtained and then the adaptive Gaussian smoothing is used to smooth the image noise region and reduce the image ambiguity. Then the contour samples of the PCB board are extracted by morphological operation and Sobel- Zernik operator, Based on in-depth analysis of the inherent characteristics of the target model, the sample points were screened, and the sample point residuals were analyzed by stratified sampling method. After repeated iterations, the best model was obtained. In the end of this paper, the algorithm is compared with the weighted least square method and the edge fitting method of Zeruike moment. It adds Gaussian noise and salt and pepper noise to the image in MAT- LAB to simulate images in different environments and uses the method of weighted least squares, edge fitting of Zernike moments and the method proposed in this paper to mark points of Mark. The results show that the method proposed in this paper based on random sample consensus algorithm PCB positioning method has higher stability and accuracy, and good robustness.

关 键 词：随机抽样一致性 标志点定位 机器视觉 椭圆拟合 

分 类 号：TN98[电子电信—信息与通信工程] TP18[自动化与计算机技术—控制理论与控制工程]